Digital Cytology and AI Applications in Cytology

const pptxgen = require("pptxgenjs");
const pres = new pptxgen();
pres.layout = "LAYOUT_16x9";
pres.title = "Digital Cytology and AI Applications";
pres.author = "Cytopathology Department";

// ─────────────────────────────────────────────
// THEME
// ─────────────────────────────────────────────
const C = {
  navy:    "0A1628",   // dominant dark
  teal:    "0D7377",   // accent 1
  sky:     "14A0A4",   // accent 2 / highlights
  mint:    "C6F1E7",   // light text/card on dark
  white:   "FFFFFF",
  offwhite:"F4F9FF",
  gray:    "5A6A7E",
  lgray:   "D4E1EF",
  red:     "E74C3C",
  gold:    "F0A500",
};

const FONT_TITLE = "Calibri";
const FONT_BODY  = "Calibri";

// ─────────────────────────────────────────────
// HELPERS
// ─────────────────────────────────────────────
function darkSlide(slide) {
  slide.addShape(pres.shapes.RECTANGLE, {
    x: 0, y: 0, w: 10, h: 5.625,
    fill: { color: C.navy }, line: { color: C.navy }
  });
}

function lightSlide(slide) {
  slide.addShape(pres.shapes.RECTANGLE, {
    x: 0, y: 0, w: 10, h: 5.625,
    fill: { color: C.offwhite }, line: { color: C.offwhite }
  });
}

function accentBar(slide, y = 0, h = 0.08, color = C.teal) {
  slide.addShape(pres.shapes.RECTANGLE, {
    x: 0, y, w: 10, h,
    fill: { color }, line: { color }
  });
}

function sectionTag(slide, text) {
  slide.addShape(pres.shapes.ROUNDED_RECTANGLE, {
    x: 0.4, y: 0.15, w: 2.2, h: 0.35,
    fill: { color: C.teal }, line: { color: C.teal }, rectRadius: 0.05
  });
  slide.addText(text.toUpperCase(), {
    x: 0.4, y: 0.15, w: 2.2, h: 0.35,
    fontSize: 8, bold: true, color: C.white,
    align: "center", valign: "middle",
    fontFace: FONT_BODY, margin: 0
  });
}

function slideNumber(slide, n) {
  slide.addText(`${n}`, {
    x: 9.6, y: 5.3, w: 0.35, h: 0.2,
    fontSize: 7, color: C.gray, align: "right", fontFace: FONT_BODY
  });
}

function addCard(slide, x, y, w, h, title, body, dark = false) {
  const bg = dark ? C.navy : C.white;
  const titleCol = dark ? C.sky : C.navy;
  const bodyCol  = dark ? C.lgray : C.gray;
  slide.addShape(pres.shapes.RECTANGLE, {
    x, y, w, h,
    fill: { color: bg },
    line: { color: C.lgray, width: 0.5 },
    shadow: { type: "outer", color: "000000", blur: 6, offset: 2, angle: 135, opacity: 0.1 }
  });
  slide.addText(title, {
    x: x + 0.15, y: y + 0.12, w: w - 0.3, h: 0.32,
    fontSize: 11, bold: true, color: titleCol, fontFace: FONT_TITLE, margin: 0
  });
  slide.addShape(pres.shapes.RECTANGLE, {
    x: x + 0.15, y: y + 0.45, w: w - 0.3, h: 0.04,
    fill: { color: C.teal }, line: { color: C.teal }
  });
  slide.addText(body, {
    x: x + 0.15, y: y + 0.55, w: w - 0.3, h: h - 0.7,
    fontSize: 9.5, color: bodyCol, fontFace: FONT_BODY,
    align: "left", valign: "top", margin: 0, wrap: true
  });
}

// ─────────────────────────────────────────────
// SLIDE 1 — TITLE
// ─────────────────────────────────────────────
{
  const sl = pres.addSlide();
  darkSlide(sl);

  // Left teal band
  sl.addShape(pres.shapes.RECTANGLE, {
    x: 0, y: 0, w: 0.18, h: 5.625,
    fill: { color: C.teal }, line: { color: C.teal }
  });

  // Bottom accent
  sl.addShape(pres.shapes.RECTANGLE, {
    x: 0, y: 5.3, w: 10, h: 0.325,
    fill: { color: C.teal }, line: { color: C.teal }
  });

  sl.addText("DIGITAL CYTOLOGY", {
    x: 0.5, y: 1.1, w: 9, h: 0.9,
    fontSize: 44, bold: true, color: C.white,
    fontFace: FONT_TITLE, align: "center", charSpacing: 3
  });
  sl.addText("& ARTIFICIAL INTELLIGENCE APPLICATIONS", {
    x: 0.5, y: 1.95, w: 9, h: 0.6,
    fontSize: 22, bold: false, color: C.sky,
    fontFace: FONT_TITLE, align: "center", charSpacing: 1
  });

  // Divider line
  sl.addShape(pres.shapes.RECTANGLE, {
    x: 2.5, y: 2.75, w: 5, h: 0.04,
    fill: { color: C.teal }, line: { color: C.teal }
  });

  sl.addText("A Comprehensive Overview for Cytopathology Practice", {
    x: 0.5, y: 3.0, w: 9, h: 0.4,
    fontSize: 14, color: C.mint, fontFace: FONT_BODY,
    align: "center", italic: true
  });

  sl.addText("From Conventional Microscopy to AI-Driven Diagnostics", {
    x: 0.5, y: 3.5, w: 9, h: 0.3,
    fontSize: 11, color: C.lgray, fontFace: FONT_BODY, align: "center"
  });

  sl.addText("June 2026", {
    x: 0.5, y: 5.3, w: 9, h: 0.25,
    fontSize: 9, color: C.white, fontFace: FONT_BODY,
    align: "center", bold: false
  });
}

// ─────────────────────────────────────────────
// SLIDE 2 — OUTLINE
// ─────────────────────────────────────────────
{
  const sl = pres.addSlide();
  lightSlide(sl);
  accentBar(sl, 0, 0.55, C.navy);
  sl.addText("PRESENTATION OUTLINE", {
    x: 0.4, y: 0, w: 9.2, h: 0.55,
    fontSize: 20, bold: true, color: C.white,
    fontFace: FONT_TITLE, valign: "middle", margin: 0
  });

  const sections = [
    ["01", "What is Digital Cytology?"],
    ["02", "The Digital Cytology Workflow"],
    ["03", "Technical Challenges"],
    ["04", "AI — Fundamentals & Architecture"],
    ["05", "Cervical Cytology: Most Mature Application"],
    ["06", "AI in Non-Gynecologic Cytology"],
    ["07", "Telecytology & Remote ROSE"],
    ["08", "Regulatory Landscape & FDA Clearance"],
    ["09", "Challenges & Barriers"],
    ["10", "Future Directions"],
  ];

  const col1 = sections.slice(0, 5);
  const col2 = sections.slice(5);

  col1.forEach(([num, label], i) => {
    const y = 0.75 + i * 0.85;
    sl.addShape(pres.shapes.RECTANGLE, {
      x: 0.4, y, w: 0.42, h: 0.42,
      fill: { color: C.teal }, line: { color: C.teal }
    });
    sl.addText(num, {
      x: 0.4, y, w: 0.42, h: 0.42,
      fontSize: 13, bold: true, color: C.white,
      align: "center", valign: "middle", fontFace: FONT_BODY, margin: 0
    });
    sl.addText(label, {
      x: 0.9, y: y + 0.04, w: 4.0, h: 0.35,
      fontSize: 12, color: C.navy, fontFace: FONT_BODY, valign: "middle"
    });
  });

  col2.forEach(([num, label], i) => {
    const y = 0.75 + i * 0.85;
    sl.addShape(pres.shapes.RECTANGLE, {
      x: 5.2, y, w: 0.42, h: 0.42,
      fill: { color: C.sky }, line: { color: C.sky }
    });
    sl.addText(num, {
      x: 5.2, y, w: 0.42, h: 0.42,
      fontSize: 13, bold: true, color: C.white,
      align: "center", valign: "middle", fontFace: FONT_BODY, margin: 0
    });
    sl.addText(label, {
      x: 5.7, y: y + 0.04, w: 4.0, h: 0.35,
      fontSize: 12, color: C.navy, fontFace: FONT_BODY, valign: "middle"
    });
  });

  slideNumber(sl, 2);
}

// ─────────────────────────────────────────────
// SLIDE 3 — WHAT IS DIGITAL CYTOLOGY
// ─────────────────────────────────────────────
{
  const sl = pres.addSlide();
  lightSlide(sl);
  accentBar(sl, 0, 0.55, C.navy);
  sectionTag(sl, "Introduction");
  sl.addText("What is Digital Cytology?", {
    x: 0.4, y: 0, w: 9.2, h: 0.55,
    fontSize: 20, bold: true, color: C.white,
    fontFace: FONT_TITLE, valign: "middle", margin: 0
  });

  // Big concept box
  sl.addShape(pres.shapes.RECTANGLE, {
    x: 0.4, y: 0.65, w: 9.2, h: 1.1,
    fill: { color: C.navy }, line: { color: C.navy }
  });
  sl.addText("Digitization of glass cytology slides → High-resolution Whole Slide Images (WSIs) accessible electronically for review, analysis, storage and remote consultation.", {
    x: 0.55, y: 0.72, w: 8.9, h: 0.95,
    fontSize: 13, color: C.mint, fontFace: FONT_BODY,
    align: "left", valign: "middle", italic: false
  });

  const pts = [
    ["Replaces conventional microscopy", "Pathologists view slides on high-resolution monitors instead of glass under a microscope"],
    ["Enables remote access", "Slides accessible from any workstation — locally or across institutions"],
    ["Foundation for AI", "Digital images are the prerequisite for any AI algorithm to analyze cytology"],
    ["Mirrors digital pathology", "Extends the principles of histopathology digitization to cytology preparations"],
  ];

  pts.forEach(([title, body], i) => {
    const x = 0.4 + (i % 2) * 4.65;
    const y = 1.85 + Math.floor(i / 2) * 1.5;
    addCard(sl, x, y, 4.5, 1.35, title, body);
  });

  slideNumber(sl, 3);
}

// ─────────────────────────────────────────────
// SLIDE 4 — CYTOLOGY vs HISTOLOGY DIGITIZATION
// ─────────────────────────────────────────────
{
  const sl = pres.addSlide();
  lightSlide(sl);
  accentBar(sl, 0, 0.55, C.navy);
  sectionTag(sl, "Introduction");
  sl.addText("Cytology vs. Histology: Why Digitization Differs", {
    x: 0.4, y: 0, w: 9.2, h: 0.55,
    fontSize: 18, bold: true, color: C.white,
    fontFace: FONT_TITLE, valign: "middle", margin: 0
  });

  // Table header
  const headers = ["Feature", "Histology", "Cytology"];
  const colX = [0.4, 3.3, 6.6];
  const colW = [2.8, 3.2, 3.2];

  headers.forEach((h, i) => {
    sl.addShape(pres.shapes.RECTANGLE, {
      x: colX[i], y: 0.65, w: colW[i], h: 0.42,
      fill: { color: C.teal }, line: { color: C.white, width: 1 }
    });
    sl.addText(h, {
      x: colX[i], y: 0.65, w: colW[i], h: 0.42,
      fontSize: 11, bold: true, color: C.white,
      align: "center", valign: "middle", fontFace: FONT_BODY, margin: 0
    });
  });

  const rows = [
    ["Slide thickness", "5 µm fixed section", "Variable — overlapping cells"],
    ["Depth of field", "Single focal plane", "Multiple focal planes needed"],
    ["Preparation types", "H&E, IHC — consistent", "Pap, MGG, H&E, Romanowsky"],
    ["Cell arrangement", "Ordered tissue architecture", "Dispersed / clustered / smeared"],
    ["Scanning speed", "Faster", "Slower — EDOF scanning needed"],
    ["AI complexity", "Moderate", "Higher — 3D cell context needed"],
  ];

  rows.forEach((row, ri) => {
    const bgColor = ri % 2 === 0 ? C.offwhite : C.white;
    row.forEach((cell, ci) => {
      sl.addShape(pres.shapes.RECTANGLE, {
        x: colX[ci], y: 1.07 + ri * 0.62, w: colW[ci], h: 0.62,
        fill: { color: bgColor }, line: { color: C.lgray, width: 0.5 }
      });
      sl.addText(cell, {
        x: colX[ci] + 0.1, y: 1.07 + ri * 0.62, w: colW[ci] - 0.2, h: 0.62,
        fontSize: 10, color: ci === 0 ? C.navy : C.gray,
        bold: ci === 0, fontFace: FONT_BODY,
        align: ci === 0 ? "left" : "center", valign: "middle", margin: 0
      });
    });
  });

  slideNumber(sl, 4);
}

// ─────────────────────────────────────────────
// SLIDE 5 — DIGITAL WORKFLOW
// ─────────────────────────────────────────────
{
  const sl = pres.addSlide();
  lightSlide(sl);
  accentBar(sl, 0, 0.55, C.navy);
  sectionTag(sl, "Workflow");
  sl.addText("The Digital Cytology Workflow", {
    x: 0.4, y: 0, w: 9.2, h: 0.55,
    fontSize: 20, bold: true, color: C.white,
    fontFace: FONT_TITLE, valign: "middle", margin: 0
  });

  // Flow boxes
  const steps = [
    { icon: "1", label: "Specimen\nCollection", sub: "FNA, LBC, smear" },
    { icon: "2", label: "Slide\nPreparation", sub: "Pap / MGG / H&E staining" },
    { icon: "3", label: "WSI\nScanning", sub: "20x–40x digital scanner" },
    { icon: "4", label: "Image\nManagement", sub: "Server / cloud storage" },
    { icon: "5", label: "AI\nAnalysis", sub: "Pre-screening & flagging" },
    { icon: "6", label: "Pathologist\nReview", sub: "Digital workstation" },
  ];

  steps.forEach((s, i) => {
    const x = 0.3 + i * 1.58;
    // Arrow (except last)
    if (i < steps.length - 1) {
      sl.addShape(pres.shapes.RECTANGLE, {
        x: x + 1.1, y: 2.05, w: 0.38, h: 0.12,
        fill: { color: C.sky }, line: { color: C.sky }
      });
      // Arrow head
      sl.addShape(pres.shapes.RIGHT_TRIANGLE, {
        x: x + 1.44, y: 1.95, w: 0.18, h: 0.32,
        fill: { color: C.sky }, line: { color: C.sky },
        rotate: 90
      });
    }
    // Circle
    sl.addShape(pres.shapes.OVAL, {
      x, y: 1.7, w: 1.1, h: 1.1,
      fill: { color: i < 3 ? C.teal : C.navy },
      line: { color: C.white, width: 2 }
    });
    sl.addText(s.icon, {
      x, y: 1.7, w: 1.1, h: 1.1,
      fontSize: 22, bold: true, color: C.white,
      align: "center", valign: "middle", fontFace: FONT_BODY, margin: 0
    });
    sl.addText(s.label, {
      x: x - 0.1, y: 2.88, w: 1.3, h: 0.55,
      fontSize: 10, bold: true, color: C.navy,
      align: "center", fontFace: FONT_BODY
    });
    sl.addText(s.sub, {
      x: x - 0.1, y: 3.42, w: 1.3, h: 0.45,
      fontSize: 8.5, color: C.gray,
      align: "center", fontFace: FONT_BODY
    });
  });

  // Key systems
  sl.addShape(pres.shapes.RECTANGLE, {
    x: 0.4, y: 4.2, w: 9.2, h: 1.2,
    fill: { color: C.navy }, line: { color: C.navy }
  });
  sl.addText("Key Systems:  Whole Slide Scanner  |  Image Management Server (IMS)  |  AI Engine  |  Review Workstation  |  LIS Integration", {
    x: 0.5, y: 4.25, w: 9.0, h: 1.1,
    fontSize: 10.5, color: C.mint, fontFace: FONT_BODY,
    align: "center", valign: "middle"
  });

  slideNumber(sl, 5);
}

// ─────────────────────────────────────────────
// SLIDE 6 — TECHNICAL CHALLENGES
// ─────────────────────────────────────────────
{
  const sl = pres.addSlide();
  lightSlide(sl);
  accentBar(sl, 0, 0.55, C.navy);
  sectionTag(sl, "Challenges");
  sl.addText("Technical Challenges in Cytology Digitization", {
    x: 0.4, y: 0, w: 9.2, h: 0.55,
    fontSize: 18, bold: true, color: C.white,
    fontFace: FONT_TITLE, valign: "middle", margin: 0
  });

  const challenges = [
    { title: "Depth of Field", body: "Cells sit at variable focal planes. EDOF (Extended Depth of Focus) scanning or z-stacking required — unlike flat 5 µm histology sections.", icon: "🔬" },
    { title: "Stain Variability", body: "Papanicolaou, MGG, H&E, Romanowsky — each has different spectral profiles. Algorithms trained on one stain may fail on another.", icon: "🎨" },
    { title: "Cell Overlap", body: "Dense cell clusters and overlapping nuclei challenge segmentation algorithms. Conventional smears are far more 3D than LBC preparations.", icon: "🔵" },
    { title: "Preparation Heterogeneity", body: "Conventional smears, ThinPrep, SurePath, cell blocks, and FNA preparations all differ — requiring separate scanner settings and AI models.", icon: "⚗️" },
    { title: "Scanning Speed", body: "Cytology slides require slower, higher-resolution scanning than histology, impacting laboratory throughput and workflow planning.", icon: "⏱️" },
    { title: "Storage & Bandwidth", body: "A single cytology WSI can be 1–5 GB. Large-scale digitization demands substantial IT infrastructure: servers, networking, and cloud capacity.", icon: "💾" },
  ];

  challenges.forEach((c, i) => {
    const x = 0.4 + (i % 3) * 3.1;
    const y = 0.7 + Math.floor(i / 3) * 2.3;
    sl.addShape(pres.shapes.RECTANGLE, {
      x, y, w: 2.9, h: 2.15,
      fill: { color: C.white },
      line: { color: C.lgray, width: 0.5 },
      shadow: { type: "outer", color: "000000", blur: 5, offset: 1, angle: 135, opacity: 0.1 }
    });
    // color top bar per card
    sl.addShape(pres.shapes.RECTANGLE, {
      x, y, w: 2.9, h: 0.32,
      fill: { color: i < 3 ? C.teal : C.sky }, line: { color: i < 3 ? C.teal : C.sky }
    });
    sl.addText(c.title, {
      x: x + 0.1, y, w: 2.7, h: 0.32,
      fontSize: 10, bold: true, color: C.white,
      fontFace: FONT_TITLE, valign: "middle", margin: 0
    });
    sl.addText(c.body, {
      x: x + 0.1, y: y + 0.38, w: 2.7, h: 1.7,
      fontSize: 9.5, color: C.gray,
      fontFace: FONT_BODY, align: "left", valign: "top", wrap: true, margin: 0
    });
  });

  slideNumber(sl, 6);
}

// ─────────────────────────────────────────────
// SLIDE 7 — AI FUNDAMENTALS
// ─────────────────────────────────────────────
{
  const sl = pres.addSlide();
  darkSlide(sl);
  accentBar(sl, 0.55, 0.04, C.teal);

  sl.addText("AI in Cytology: Core Concepts", {
    x: 0.4, y: 0.08, w: 9.2, h: 0.5,
    fontSize: 22, bold: true, color: C.white,
    fontFace: FONT_TITLE, valign: "middle"
  });

  sectionTag(sl, "AI Fundamentals");

  const models = [
    { name: "CNN", full: "Convolutional Neural Network", desc: "The workhorse of image AI. Learns hierarchical features from pixels. Excellent at single-cell classification. Dominant in most cytology AI papers.", color: C.teal },
    { name: "ViT", full: "Vision Transformer", desc: "Captures global context across the whole slide. Identifies spatial relationships between cells — important for architectural features like clusters.", color: C.sky },
    { name: "CNN + ViT", full: "Hybrid Architecture", desc: "Best of both worlds. Local feature extraction (CNN) + global context (ViT). Currently the highest-performing approach in cervical cytology.", color: C.gold },
    { name: "MIL", full: "Multi-Instance Learning", desc: "Trains on slide-level labels without needing per-cell annotation. Practical for large datasets where detailed labeling is expensive.", color: "8A7FD4" },
    { name: "ANN", full: "Artificial Neural Network", desc: "Integrates structured clinical data (age, HPV status, Bethesda category) with image features for richer, multimodal diagnostic outputs.", color: "D4766A" },
    { name: "FM", full: "Foundation Model", desc: "Pre-trained on millions of WSIs. Fine-tuned for specific tasks. Reduces need for large labeled datasets. Emerging frontier in 2025–2026.", color: "7AD46A" },
  ];

  models.forEach((m, i) => {
    const x = 0.3 + (i % 3) * 3.2;
    const y = 0.75 + Math.floor(i / 3) * 2.25;
    sl.addShape(pres.shapes.RECTANGLE, {
      x, y, w: 3.0, h: 2.1,
      fill: { color: "0D2140" }, line: { color: m.color, width: 1.5 }
    });
    sl.addText(m.name, {
      x: x + 0.12, y: y + 0.1, w: 1.0, h: 0.38,
      fontSize: 16, bold: true, color: m.color,
      fontFace: FONT_TITLE, margin: 0
    });
    sl.addText(m.full, {
      x: x + 0.12, y: y + 0.48, w: 2.75, h: 0.28,
      fontSize: 9, color: C.lgray, fontFace: FONT_BODY,
      italic: true, margin: 0
    });
    sl.addShape(pres.shapes.RECTANGLE, {
      x: x + 0.12, y: y + 0.78, w: 2.75, h: 0.03,
      fill: { color: m.color }, line: { color: m.color }
    });
    sl.addText(m.desc, {
      x: x + 0.12, y: y + 0.86, w: 2.75, h: 1.1,
      fontSize: 9, color: C.lgray,
      fontFace: FONT_BODY, wrap: true, valign: "top", margin: 0
    });
  });

  slideNumber(sl, 7);
}

// ─────────────────────────────────────────────
// SLIDE 8 — CERVICAL CYTOLOGY (Gynecologic AI)
// ─────────────────────────────────────────────
{
  const sl = pres.addSlide();
  lightSlide(sl);
  accentBar(sl, 0, 0.55, C.navy);
  sectionTag(sl, "Gynecologic Cytology");
  sl.addText("Cervical Cytology: The Most Mature AI Application", {
    x: 0.4, y: 0, w: 9.2, h: 0.55,
    fontSize: 18, bold: true, color: C.white,
    fontFace: FONT_TITLE, valign: "middle", margin: 0
  });

  // Timeline
  const timeline = [
    { yr: "1990s", ev: "Computer-assisted detection (AutoPap, FocalPoint) — first generation" },
    { yr: "2000s", ev: "ThinPrep Imaging System — AI-guided slide ranking commercially deployed" },
    { yr: "2020s", ev: "Deep learning CNNs surpass prior systems in sensitivity & specificity" },
    { yr: "2023+", ev: "Foundation models, hybrid CNN-ViT, slide-level classification" },
    { yr: "2024", ev: "Hologic Genius — first & only FDA-cleared digital cytology system" },
  ];

  timeline.forEach((t, i) => {
    sl.addShape(pres.shapes.OVAL, {
      x: 0.55, y: 0.72 + i * 0.82, w: 0.72, h: 0.42,
      fill: { color: C.teal }, line: { color: C.teal }
    });
    sl.addText(t.yr, {
      x: 0.55, y: 0.72 + i * 0.82, w: 0.72, h: 0.42,
      fontSize: 8.5, bold: true, color: C.white,
      align: "center", valign: "middle", fontFace: FONT_BODY, margin: 0
    });
    sl.addText(t.ev, {
      x: 1.4, y: 0.76 + i * 0.82, w: 4.4, h: 0.38,
      fontSize: 10, color: C.navy, fontFace: FONT_BODY, valign: "middle"
    });
    if (i < timeline.length - 1) {
      sl.addShape(pres.shapes.RECTANGLE, {
        x: 0.85, y: 1.14 + i * 0.82, w: 0.08, h: 0.4,
        fill: { color: C.lgray }, line: { color: C.lgray }
      });
    }
  });

  // Stats box
  sl.addShape(pres.shapes.RECTANGLE, {
    x: 6.0, y: 0.65, w: 3.6, h: 4.7,
    fill: { color: C.navy }, line: { color: C.navy }
  });
  sl.addText("HOLOGIC GENIUS SYSTEM", {
    x: 6.05, y: 0.72, w: 3.5, h: 0.38,
    fontSize: 10, bold: true, color: C.sky, align: "center",
    fontFace: FONT_TITLE, charSpacing: 1, margin: 0
  });
  sl.addShape(pres.shapes.RECTANGLE, {
    x: 6.2, y: 1.12, w: 3.2, h: 0.04,
    fill: { color: C.teal }, line: { color: C.teal }
  });

  const stats = [
    ["50–60%", "Reduction in review time vs. manual ThinPrep"],
    ["FDA", "Only cleared digital cytology system in the USA"],
    ["3D", "Volumetric imaging technology for LBC slides"],
    ["AI-guided", "Flags pre-cancerous & malignant cells for review"],
  ];

  stats.forEach((s, i) => {
    sl.addText(s[0], {
      x: 6.1, y: 1.25 + i * 0.98, w: 3.4, h: 0.45,
      fontSize: 22, bold: true, color: C.gold,
      align: "center", fontFace: FONT_TITLE, margin: 0
    });
    sl.addText(s[1], {
      x: 6.1, y: 1.68 + i * 0.98, w: 3.4, h: 0.38,
      fontSize: 9, color: C.lgray, align: "center",
      fontFace: FONT_BODY, margin: 0
    });
  });

  slideNumber(sl, 8);
}

// ─────────────────────────────────────────────
// SLIDE 9 — THYROID CYTOLOGY AI
// ─────────────────────────────────────────────
{
  const sl = pres.addSlide();
  lightSlide(sl);
  accentBar(sl, 0, 0.55, C.navy);
  sectionTag(sl, "Non-Gynecologic AI");
  sl.addText("AI in Thyroid Cytology (FNA)", {
    x: 0.4, y: 0, w: 9.2, h: 0.55,
    fontSize: 20, bold: true, color: C.white,
    fontFace: FONT_TITLE, valign: "middle", margin: 0
  });

  // Bethesda box
  sl.addShape(pres.shapes.RECTANGLE, {
    x: 0.4, y: 0.65, w: 4.3, h: 4.7,
    fill: { color: C.navy }, line: { color: C.navy }
  });
  sl.addText("BETHESDA SYSTEM", {
    x: 0.5, y: 0.72, w: 4.1, h: 0.35,
    fontSize: 11, bold: true, color: C.sky,
    align: "center", fontFace: FONT_TITLE, charSpacing: 2, margin: 0
  });

  const bethesda = [
    ["I", "Non-diagnostic", "—"],
    ["II", "Benign", "~2%"],
    ["III", "AUS / FLUS", "10–30%"],
    ["IV", "FN / SFN", "25–40%"],
    ["V", "Suspicious malignancy", "60–75%"],
    ["VI", "Malignant", ">97%"],
  ];

  const bColors = ["888888", "27AE60", C.red, C.gold, "E67E22", C.red];

  bethesda.forEach((b, i) => {
    const by = 1.15 + i * 0.67;
    sl.addShape(pres.shapes.RECTANGLE, {
      x: 0.55, y: by, w: 0.38, h: 0.48,
      fill: { color: bColors[i] }, line: { color: bColors[i] }
    });
    sl.addText(b[0], {
      x: 0.55, y: by, w: 0.38, h: 0.48,
      fontSize: 12, bold: true, color: C.white,
      align: "center", valign: "middle", fontFace: FONT_BODY, margin: 0
    });
    sl.addText(b[1], {
      x: 1.02, y: by + 0.06, w: 2.4, h: 0.35,
      fontSize: 10, color: C.lgray, fontFace: FONT_BODY
    });
    sl.addText(b[2], {
      x: 3.55, y: by + 0.06, w: 0.9, h: 0.35,
      fontSize: 9, color: C.gold, fontFace: FONT_BODY, bold: true
    });
  });

  sl.addShape(pres.shapes.RECTANGLE, {
    x: 3.5, y: 1.0, w: 0.95, h: 0.18,
    fill: { color: C.navy }, line: { color: C.navy }
  });
  sl.addText("Malig. Risk", {
    x: 3.5, y: 1.0, w: 0.95, h: 0.18,
    fontSize: 7.5, color: C.gray, fontFace: FONT_BODY, margin: 0
  });

  // AI for AUS/FLUS
  sl.addShape(pres.shapes.RECTANGLE, {
    x: 4.85, y: 0.65, w: 4.75, h: 4.7,
    fill: { color: C.offwhite }, line: { color: C.lgray, width: 0.5 }
  });
  sl.addText("AI Targets the Indeterminate Zone", {
    x: 5.0, y: 0.72, w: 4.45, h: 0.38,
    fontSize: 12, bold: true, color: C.navy,
    fontFace: FONT_TITLE, margin: 0
  });
  sl.addShape(pres.shapes.RECTANGLE, {
    x: 5.0, y: 1.12, w: 4.25, h: 0.04,
    fill: { color: C.teal }, line: { color: C.teal }
  });

  const aipts = [
    "AUS/FLUS carries 10–30% malignancy risk — highly uncertain",
    "13 AI studies reviewed (Poursina et al., Acta Cytol 2025)",
    "CNN excels at WSI feature extraction",
    "ANN integrates clinical + image data",
    "Combined CNN+ANN = best diagnostic accuracy",
    "Classifiers: YOLOv4, EfficientNetV2, MobileNetV2",
    "Goal: reduce unnecessary surgery in benign AUS/FLUS cases",
    "Challenge: No standardization across scanner platforms yet",
  ];

  sl.addText(aipts.map(p => ({ text: p, options: { bullet: true, breakLine: true } })), {
    x: 5.0, y: 1.25, w: 4.45, h: 4.0,
    fontSize: 10, color: C.gray, fontFace: FONT_BODY,
    align: "left", valign: "top"
  });

  slideNumber(sl, 9);
}

// ─────────────────────────────────────────────
// SLIDE 10 — URINE & OTHER NON-GYN CYTOLOGY
// ─────────────────────────────────────────────
{
  const sl = pres.addSlide();
  lightSlide(sl);
  accentBar(sl, 0, 0.55, C.navy);
  sectionTag(sl, "Non-Gynecologic AI");
  sl.addText("AI in Non-Gynecologic Cytology", {
    x: 0.4, y: 0, w: 9.2, h: 0.55,
    fontSize: 20, bold: true, color: C.white,
    fontFace: FONT_TITLE, valign: "middle", margin: 0
  });

  const organs = [
    {
      organ: "Urine Cytology",
      color: C.teal,
      points: [
        "Paris System classification",
        "11 AI studies reviewed (Nabiyouni & Chiou, AJCP 2026)",
        "Sensitivity 63–100% for HGUC detection",
        "Specificity 61.8–100%",
        "Predominantly CNN-based models",
        "Reduces false negatives in high-grade urothelial carcinoma"
      ]
    },
    {
      organ: "Pulmonary Cytology",
      color: C.sky,
      points: [
        "EBUS-guided FNA and BAL specimens",
        "Deep learning for lung cancer subtype classification",
        "Adenocarcinoma vs. SCC vs. SCLC vs. metastasis",
        "AI-assisted ROSE during bronchoscopy",
        "Molecular prediction from morphology (EGFR, ALK)"
      ]
    },
    {
      organ: "Other Sites",
      color: C.navy,
      points: [
        "Breast FNA: benign vs. atypical vs. malignant",
        "Serous fluids: malignant cell detection",
        "Salivary gland FNA: classification",
        "Lymph node FNA: lymphoma vs. reactive",
        "Body fluid differential cell counts (automated)"
      ]
    },
  ];

  organs.forEach((o, i) => {
    const y = 0.65 + i * 1.6;
    sl.addShape(pres.shapes.RECTANGLE, {
      x: 0.4, y, w: 9.2, h: 1.5,
      fill: { color: C.white }, line: { color: C.lgray, width: 0.5 }
    });
    sl.addShape(pres.shapes.RECTANGLE, {
      x: 0.4, y, w: 2.2, h: 1.5,
      fill: { color: o.color }, line: { color: o.color }
    });
    sl.addText(o.organ, {
      x: 0.4, y, w: 2.2, h: 1.5,
      fontSize: 12, bold: true, color: C.white,
      align: "center", valign: "middle", fontFace: FONT_TITLE,
      wrap: true, margin: 0
    });
    sl.addText(o.points.map(p => ({ text: p, options: { bullet: true, breakLine: true } })), {
      x: 2.75, y: y + 0.05, w: 6.7, h: 1.4,
      fontSize: 9.5, color: C.gray, fontFace: FONT_BODY,
      align: "left", valign: "top"
    });
  });

  slideNumber(sl, 10);
}

// ─────────────────────────────────────────────
// SLIDE 11 — TELECYTOLOGY & ROSE
// ─────────────────────────────────────────────
{
  const sl = pres.addSlide();
  darkSlide(sl);
  accentBar(sl, 0.55, 0.04, C.teal);

  sl.addText("Telecytology & Remote ROSE", {
    x: 0.4, y: 0.08, w: 7, h: 0.5,
    fontSize: 22, bold: true, color: C.white,
    fontFace: FONT_TITLE, valign: "middle"
  });
  sectionTag(sl, "Telecytology");

  // ROSE explanation
  sl.addShape(pres.shapes.RECTANGLE, {
    x: 0.4, y: 0.68, w: 9.2, h: 0.9,
    fill: { color: "0D2140" }, line: { color: C.sky, width: 1 }
  });
  sl.addText("ROSE = Rapid On-Site Evaluation — real-time adequacy assessment of FNA specimens DURING the procedure, traditionally requiring an on-site cytopathologist.", {
    x: 0.55, y: 0.72, w: 8.9, h: 0.82,
    fontSize: 11.5, color: C.mint, fontFace: FONT_BODY, valign: "middle"
  });

  const teleCards = [
    { title: "Traditional ROSE", body: "Cytotechnologist or pathologist physically present at procedure suite. Requires scheduling, travel, dedicated personnel — not feasible in all centers.", accent: C.gray },
    { title: "Telecytology ROSE", body: "Digital images transmitted in real-time from procedure room to remote pathologist. Enables expert consultation across geography without physical presence.", accent: C.teal },
    { title: "AI-Assisted ROSE", body: "Automated algorithms assess specimen adequacy instantly. Alerts staff if sample is unsatisfactory. Reduces unnecessary repeat procedures.", accent: C.sky },
    { title: "Global Equity", body: "Telecytology democratizes expert cytopathology. Rural, low-resource, and remote settings gain access to subspecialist opinion without specialist relocation.", accent: C.gold },
  ];

  teleCards.forEach((c, i) => {
    const x = 0.4 + (i % 2) * 4.8;
    const y = 1.7 + Math.floor(i / 2) * 1.85;
    sl.addShape(pres.shapes.RECTANGLE, {
      x, y, w: 4.55, h: 1.7,
      fill: { color: "0D2140" }, line: { color: c.accent, width: 1 }
    });
    sl.addShape(pres.shapes.RECTANGLE, {
      x, y, w: 4.55, h: 0.04,
      fill: { color: c.accent }, line: { color: c.accent }
    });
    sl.addText(c.title, {
      x: x + 0.15, y: y + 0.1, w: 4.25, h: 0.35,
      fontSize: 12, bold: true, color: c.accent,
      fontFace: FONT_TITLE, margin: 0
    });
    sl.addText(c.body, {
      x: x + 0.15, y: y + 0.5, w: 4.25, h: 1.1,
      fontSize: 9.5, color: C.lgray, fontFace: FONT_BODY,
      valign: "top", wrap: true, margin: 0
    });
  });

  slideNumber(sl, 11);
}

// ─────────────────────────────────────────────
// SLIDE 12 — BEYOND DETECTION: AI APPLICATIONS
// ─────────────────────────────────────────────
{
  const sl = pres.addSlide();
  lightSlide(sl);
  accentBar(sl, 0, 0.55, C.navy);
  sectionTag(sl, "AI Applications");
  sl.addText("AI Beyond Detection: Extended Applications", {
    x: 0.4, y: 0, w: 9.2, h: 0.55,
    fontSize: 18, bold: true, color: C.white,
    fontFace: FONT_TITLE, valign: "middle", margin: 0
  });

  const apps = [
    { title: "Biomarker Quantification", body: "Automated scoring of Ki-67, p16, HPV-related markers on cytology preparations. Removes inter-observer variability in manual counting.", num: "01" },
    { title: "Molecular Prediction", body: "Predicting BRAF/RAS/EGFR mutation status from morphology alone — without expensive molecular testing.", num: "02" },
    { title: "Prognosis", body: "Predicting clinical outcomes (recurrence, survival) directly from cytologic appearance of cancer cells.", num: "03" },
    { title: "Quality Control", body: "Automated detection of poor staining, obscuring blood, insufficient cellularity, and scanner artefacts.", num: "04" },
    { title: "Screening Triage", body: "Ranking slides by abnormality likelihood — directing cytotechnologist attention to highest-risk cases first.", num: "05" },
    { title: "Cell Counting", body: "Automated differential counts in body fluids and CSF — faster and more reproducible than manual methods.", num: "06" },
  ];

  apps.forEach((a, i) => {
    const x = 0.4 + (i % 3) * 3.1;
    const y = 0.7 + Math.floor(i / 3) * 2.35;
    sl.addShape(pres.shapes.RECTANGLE, {
      x, y, w: 2.9, h: 2.2,
      fill: { color: C.white }, line: { color: C.lgray, width: 0.5 },
      shadow: { type: "outer", color: "000000", blur: 5, offset: 1, angle: 135, opacity: 0.1 }
    });
    sl.addShape(pres.shapes.OVAL, {
      x: x + 0.15, y: y + 0.15, w: 0.5, h: 0.5,
      fill: { color: C.teal }, line: { color: C.teal }
    });
    sl.addText(a.num, {
      x: x + 0.15, y: y + 0.15, w: 0.5, h: 0.5,
      fontSize: 11, bold: true, color: C.white,
      align: "center", valign: "middle", fontFace: FONT_BODY, margin: 0
    });
    sl.addText(a.title, {
      x: x + 0.75, y: y + 0.18, w: 2.05, h: 0.45,
      fontSize: 10.5, bold: true, color: C.navy,
      fontFace: FONT_TITLE, valign: "middle", margin: 0
    });
    sl.addShape(pres.shapes.RECTANGLE, {
      x: x + 0.15, y: y + 0.72, w: 2.6, h: 0.03,
      fill: { color: C.lgray }, line: { color: C.lgray }
    });
    sl.addText(a.body, {
      x: x + 0.15, y: y + 0.82, w: 2.6, h: 1.28,
      fontSize: 9.5, color: C.gray, fontFace: FONT_BODY,
      valign: "top", wrap: true, margin: 0
    });
  });

  slideNumber(sl, 12);
}

// ─────────────────────────────────────────────
// SLIDE 13 — REGULATORY LANDSCAPE
// ─────────────────────────────────────────────
{
  const sl = pres.addSlide();
  lightSlide(sl);
  accentBar(sl, 0, 0.55, C.navy);
  sectionTag(sl, "Regulation");
  sl.addText("Regulatory Landscape & FDA Clearance", {
    x: 0.4, y: 0, w: 9.2, h: 0.55,
    fontSize: 18, bold: true, color: C.white,
    fontFace: FONT_TITLE, valign: "middle", margin: 0
  });

  // Key stat
  sl.addShape(pres.shapes.RECTANGLE, {
    x: 0.4, y: 0.68, w: 9.2, h: 1.05,
    fill: { color: C.navy }, line: { color: C.navy }
  });
  sl.addText([
    { text: "Only 3 ", options: { bold: true, color: C.gold, fontSize: 20 } },
    { text: "AI/ML Software as Medical Devices in digital pathology have received FDA clearance as of 2026", options: { bold: false, color: C.mint, fontSize: 14 } },
    { text: "   —   The gap is VALIDATION DATASETS, not regulatory pathways", options: { bold: true, color: C.sky, fontSize: 13 } }
  ], {
    x: 0.55, y: 0.72, w: 8.9, h: 0.95,
    fontFace: FONT_BODY, valign: "middle"
  });

  const regs = [
    { title: "Hologic Genius", body: "FDA-cleared for primary cervical cytology screening. The only approved system in the USA. Uses volumetric imaging + AI.", tag: "FDA CLEARED", tagC: "27AE60" },
    { title: "ASC Task Force (2024)", body: "American Society of Cytopathology published best-practice recommendations for AI validation in cytology — Kim et al., JASC 2024.", tag: "GUIDELINES", tagC: C.sky },
    { title: "DICOM Standards", body: "WSI supplement of DICOM enables interoperability between scanners, AI platforms, and archive systems. Critical for multi-site validation.", tag: "STANDARD", tagC: C.teal },
    { title: "NCI Workshop (2024)", body: "National Cancer Institute convened experts to address data standardization, de-identification, AI validation, and cloud-based infrastructure.", tag: "POLICY", tagC: C.gold },
  ];

  regs.forEach((r, i) => {
    const x = 0.4 + (i % 2) * 4.8;
    const y = 1.85 + Math.floor(i / 2) * 1.75;
    sl.addShape(pres.shapes.RECTANGLE, {
      x, y, w: 4.55, h: 1.6,
      fill: { color: C.white }, line: { color: C.lgray, width: 0.5 },
      shadow: { type: "outer", color: "000000", blur: 4, offset: 1, angle: 135, opacity: 0.1 }
    });
    sl.addShape(pres.shapes.ROUNDED_RECTANGLE, {
      x: x + 0.15, y: y + 0.12, w: 1.2, h: 0.28,
      fill: { color: r.tagC }, line: { color: r.tagC }, rectRadius: 0.04
    });
    sl.addText(r.tag, {
      x: x + 0.15, y: y + 0.12, w: 1.2, h: 0.28,
      fontSize: 7, bold: true, color: C.white,
      align: "center", valign: "middle", fontFace: FONT_BODY, margin: 0
    });
    sl.addText(r.title, {
      x: x + 1.45, y: y + 0.1, w: 2.9, h: 0.32,
      fontSize: 11, bold: true, color: C.navy,
      fontFace: FONT_TITLE, valign: "middle", margin: 0
    });
    sl.addText(r.body, {
      x: x + 0.15, y: y + 0.5, w: 4.25, h: 1.0,
      fontSize: 9.5, color: C.gray, fontFace: FONT_BODY,
      valign: "top", wrap: true, margin: 0
    });
  });

  slideNumber(sl, 13);
}

// ─────────────────────────────────────────────
// SLIDE 14 — CHALLENGES & BARRIERS
// ─────────────────────────────────────────────
{
  const sl = pres.addSlide();
  darkSlide(sl);
  accentBar(sl, 0.55, 0.04, C.teal);

  sl.addText("Challenges & Barriers to Implementation", {
    x: 0.4, y: 0.08, w: 9, h: 0.5,
    fontSize: 21, bold: true, color: C.white,
    fontFace: FONT_TITLE, valign: "middle"
  });
  sectionTag(sl, "Challenges");

  const barriers = [
    { cat: "Technical", items: ["Algorithm bias from non-representative training data", "Scanner and stain variability across institutions", "Lack of standardized validation frameworks", "Data storage and bandwidth demands (1–5 GB per WSI)"], col: C.teal },
    { cat: "Clinical", items: ["Black-box AI — limited explainability for clinicians", "No peer-reviewed real-world validation for most non-gyn AI", "Risk of over-reliance on AI in ambiguous cases", "Integration with existing LIS/PACS workflows"], col: C.sky },
    { cat: "Legal & Ethical", items: ["Unclear medicolegal liability when AI contributes to error", "Patient data privacy in cloud-based AI platforms", "IP ownership of AI algorithms trained on patient slides", "Informed consent for AI analysis of specimens"], col: C.gold },
    { cat: "Financial", items: ["High upfront scanner and server costs", "No established CPT reimbursement for AI-assisted cytology", "Ongoing maintenance and software licensing fees", "Training and change management for lab staff"], col: "D4766A" },
  ];

  barriers.forEach((b, i) => {
    const x = 0.3 + (i % 2) * 4.8;
    const y = 0.72 + Math.floor(i / 2) * 2.3;
    sl.addShape(pres.shapes.RECTANGLE, {
      x, y, w: 4.55, h: 2.15,
      fill: { color: "0D2140" }, line: { color: b.col, width: 1.5 }
    });
    sl.addText(b.cat, {
      x: x + 0.15, y: y + 0.1, w: 4.2, h: 0.34,
      fontSize: 12, bold: true, color: b.col,
      fontFace: FONT_TITLE, margin: 0
    });
    sl.addText(b.items.map(it => ({ text: it, options: { bullet: true, breakLine: true } })), {
      x: x + 0.15, y: y + 0.5, w: 4.2, h: 1.55,
      fontSize: 9.5, color: C.lgray, fontFace: FONT_BODY,
      valign: "top"
    });
  });

  slideNumber(sl, 14);
}

// ─────────────────────────────────────────────
// SLIDE 15 — GLOBAL & EQUITY
// ─────────────────────────────────────────────
{
  const sl = pres.addSlide();
  lightSlide(sl);
  accentBar(sl, 0, 0.55, C.navy);
  sectionTag(sl, "Global Health");
  sl.addText("Global Health & Equity Implications", {
    x: 0.4, y: 0, w: 9.2, h: 0.55,
    fontSize: 20, bold: true, color: C.white,
    fontFace: FONT_TITLE, valign: "middle", margin: 0
  });

  // Global burden callout
  sl.addShape(pres.shapes.RECTANGLE, {
    x: 0.4, y: 0.68, w: 9.2, h: 0.88,
    fill: { color: C.teal }, line: { color: C.teal }
  });
  sl.addText("90% of cervical cancer deaths occur in low- and middle-income countries — where pathologist density is lowest.", {
    x: 0.55, y: 0.72, w: 8.9, h: 0.8,
    fontSize: 13, bold: true, color: C.white,
    fontFace: FONT_BODY, align: "center", valign: "middle"
  });

  const potentials = [
    { hd: "Scalable Screening", body: "AI enables high-volume cervical cytology screening without requiring proportional pathologist growth." },
    { hd: "Telecytology Access", body: "Remote expert review via telecytology brings subspecialist-level cytopathology to resource-limited settings." },
    { hd: "Reduced Variability", body: "AI removes inter-observer variation in settings with less experienced personnel or high staff turnover." },
    { hd: "Cloud-Based Solutions", body: "Cloud platforms reduce on-site IT burden — slides stored and AI processed remotely with local-only review interface." },
  ];

  potentials.forEach((p, i) => {
    const x = 0.4 + (i % 2) * 4.8;
    const y = 1.7 + Math.floor(i / 2) * 1.75;
    sl.addShape(pres.shapes.RECTANGLE, {
      x, y, w: 4.55, h: 1.6,
      fill: { color: C.white }, line: { color: C.lgray, width: 0.5 }
    });
    sl.addShape(pres.shapes.RECTANGLE, {
      x, y, w: 0.08, h: 1.6,
      fill: { color: C.teal }, line: { color: C.teal }
    });
    sl.addText(p.hd, {
      x: x + 0.22, y: y + 0.12, w: 4.2, h: 0.35,
      fontSize: 12, bold: true, color: C.navy,
      fontFace: FONT_TITLE, margin: 0
    });
    sl.addText(p.body, {
      x: x + 0.22, y: y + 0.5, w: 4.2, h: 1.0,
      fontSize: 10, color: C.gray, fontFace: FONT_BODY,
      valign: "top", wrap: true, margin: 0
    });
  });

  slideNumber(sl, 15);
}

// ─────────────────────────────────────────────
// SLIDE 16 — FUTURE DIRECTIONS
// ─────────────────────────────────────────────
{
  const sl = pres.addSlide();
  lightSlide(sl);
  accentBar(sl, 0, 0.55, C.navy);
  sectionTag(sl, "Future");
  sl.addText("Future Directions", {
    x: 0.4, y: 0, w: 9.2, h: 0.55,
    fontSize: 22, bold: true, color: C.white,
    fontFace: FONT_TITLE, valign: "middle", margin: 0
  });

  const futureData = [
    { area: "Cervical Cytology", current: "FDA-cleared, commercially deployed", future: "Fully automated primary screening — pathologist confirms AI calls" },
    { area: "Thyroid FNA", current: "Research / pilot stage", future: "AI triage of Bethesda III/IV to replace or complement molecular tests" },
    { area: "Urine Cytology", current: "Research / pilot stage", future: "AI-integrated Paris System reporting with HGUC confidence score" },
    { area: "Pulmonary Cytology", current: "Emerging", future: "ROSE automation + real-time molecular marker prediction" },
    { area: "Body Fluid AI", current: "Emerging", future: "Automated differential + malignancy likelihood reporting" },
    { area: "Foundation Models", current: "Early deployment (2025–26)", future: "Universal pre-trained cytology model adaptable to all specimen types" },
  ];

  // Table
  const headers2 = ["Cytology Area", "Current Status", "Future Prospect"];
  const colX2 = [0.4, 3.0, 6.5];
  const colW2 = [2.55, 3.45, 3.1];

  headers2.forEach((h, i) => {
    sl.addShape(pres.shapes.RECTANGLE, {
      x: colX2[i], y: 0.65, w: colW2[i], h: 0.4,
      fill: { color: C.teal }, line: { color: C.white, width: 0.5 }
    });
    sl.addText(h, {
      x: colX2[i], y: 0.65, w: colW2[i], h: 0.4,
      fontSize: 10, bold: true, color: C.white,
      align: "center", valign: "middle", fontFace: FONT_BODY, margin: 0
    });
  });

  futureData.forEach((row, ri) => {
    const cells = [row.area, row.current, row.future];
    const rowBg = ri % 2 === 0 ? C.offwhite : C.white;
    cells.forEach((cell, ci) => {
      sl.addShape(pres.shapes.RECTANGLE, {
        x: colX2[ci], y: 1.05 + ri * 0.75, w: colW2[ci], h: 0.75,
        fill: { color: rowBg }, line: { color: C.lgray, width: 0.5 }
      });
      sl.addText(cell, {
        x: colX2[ci] + 0.1, y: 1.05 + ri * 0.75, w: colW2[ci] - 0.2, h: 0.75,
        fontSize: 9.5, color: ci === 0 ? C.navy : C.gray,
        bold: ci === 0, fontFace: FONT_BODY,
        align: ci === 0 ? "left" : "left", valign: "middle",
        wrap: true, margin: 0
      });
    });
  });

  slideNumber(sl, 16);
}

// ─────────────────────────────────────────────
// SLIDE 17 — KEY TAKEAWAYS
// ─────────────────────────────────────────────
{
  const sl = pres.addSlide();
  darkSlide(sl);
  accentBar(sl, 0.55, 0.04, C.teal);

  sl.addText("Key Takeaways", {
    x: 0.4, y: 0.08, w: 9, h: 0.5,
    fontSize: 24, bold: true, color: C.white,
    fontFace: FONT_TITLE, valign: "middle"
  });

  const takes = [
    { n: "1", msg: "Digital cytology converts glass slides to WSIs — the essential prerequisite for AI analysis and telecytology." },
    { n: "2", msg: "Cytology digitization is harder than histology: overlapping cells, variable focal planes, multiple stain types." },
    { n: "3", msg: "Hologic Genius is the only FDA-cleared AI cytology system. All others are research or investigational." },
    { n: "4", msg: "AI is most mature in cervical cytology (30+ years). Non-gynecologic applications (thyroid, urine, lung) are emerging rapidly." },
    { n: "5", msg: "Beyond detection: AI quantifies biomarkers, predicts molecular mutations, controls quality, and automates cell counts." },
    { n: "6", msg: "Major barriers: bias, explainability, liability, reimbursement, and lack of standardized validation." },
    { n: "7", msg: "AI + telecytology could transform global equity in cervical cancer screening — especially in low-resource settings." },
  ];

  takes.forEach((t, i) => {
    sl.addShape(pres.shapes.OVAL, {
      x: 0.45, y: 0.72 + i * 0.7, w: 0.4, h: 0.4,
      fill: { color: C.teal }, line: { color: C.teal }
    });
    sl.addText(t.n, {
      x: 0.45, y: 0.72 + i * 0.7, w: 0.4, h: 0.4,
      fontSize: 12, bold: true, color: C.white,
      align: "center", valign: "middle", fontFace: FONT_BODY, margin: 0
    });
    sl.addText(t.msg, {
      x: 1.0, y: 0.75 + i * 0.7, w: 8.6, h: 0.38,
      fontSize: 10.5, color: i % 2 === 0 ? C.lgray : C.mint,
      fontFace: FONT_BODY, valign: "middle"
    });
  });

  slideNumber(sl, 17);
}

// ─────────────────────────────────────────────
// SLIDE 18 — REFERENCES & CLOSE
// ─────────────────────────────────────────────
{
  const sl = pres.addSlide();
  darkSlide(sl);

  // Left teal band
  sl.addShape(pres.shapes.RECTANGLE, {
    x: 0, y: 0, w: 0.18, h: 5.625,
    fill: { color: C.teal }, line: { color: C.teal }
  });

  sl.addText("References & Sources", {
    x: 0.5, y: 0.15, w: 9.1, h: 0.45,
    fontSize: 18, bold: true, color: C.sky,
    fontFace: FONT_TITLE
  });
  sl.addShape(pres.shapes.RECTANGLE, {
    x: 0.5, y: 0.6, w: 9, h: 0.04,
    fill: { color: C.teal }, line: { color: C.teal }
  });

  const refs = [
    "1. Satturwar S, Parwani AV, Li Z. Advances in Digital Cytopathology and Artificial Intelligence Applications. Adv Anat Pathol. 2026. PMID 42046894",
    "2. Kim D, et al. Digital cytology part 1: implementation for practice — ASC Task Force. J Am Soc Cytopathol. 2024. PMID 38158316",
    "3. Kim D, et al. Digital cytology part 2: AI in cytology — ASC Task Force. J Am Soc Cytopathol. 2024. PMID 38158317",
    "4. Poursina O, et al. AI and WSI Assist in Thyroid Indeterminate Cytology: Systematic Review. Acta Cytol. 2025. PMID 39746329",
    "5. Nabiyouni F, Chiou PZ. Enhancing urine cytopathology with AI: systematic review. Am J Clin Pathol. 2026. PMID 41643205",
    "6. Makhlouf HR, et al. Digital Pathology Imaging AI in Cancer Research: NCI Workshop Report. J Pathol Inform. 2026. PMID 41476571",
    "7. Valles-Coral et al. AI in Cervical Cytology — Systematic Review. Frontiers Big Data. 2026",
    "8. Bailey & Love's Short Practice of Surgery, 28th Ed. — Histology & Cytology sections, p. 202–204",
  ];

  sl.addText(refs.map(r => ({ text: r, options: { breakLine: true } })), {
    x: 0.5, y: 0.72, w: 9.1, h: 3.6,
    fontSize: 9, color: C.lgray, fontFace: FONT_BODY,
    valign: "top", lineSpacingMultiple: 1.3
  });

  // Closing bar
  sl.addShape(pres.shapes.RECTANGLE, {
    x: 0, y: 4.9, w: 10, h: 0.725,
    fill: { color: C.teal }, line: { color: C.teal }
  });
  sl.addText("Thank You  |  Questions Welcome", {
    x: 0, y: 4.9, w: 10, h: 0.725,
    fontSize: 16, bold: true, color: C.white,
    align: "center", valign: "middle", fontFace: FONT_TITLE, charSpacing: 2
  });

  slideNumber(sl, 18);
}

// ─────────────────────────────────────────────
// WRITE FILE
// ─────────────────────────────────────────────
pres.writeFile({ fileName: "/home/daytona/workspace/digital-cytology-ppt/Digital_Cytology_and_AI.pptx" })
  .then(() => console.log("✅ Presentation saved successfully."))
  .catch(err => console.error("❌ Error:", err));