mkdir -p /tmp/workspace/ent-pg-answers && ls /tmp/workspace/ent-pg-answers/

cat << 'PYEOF' > /tmp/workspace/ent-pg-answers/create_doc.py from docx import Document from docx.shared import Pt, Inches, RGBColor from docx.enum.text import WD_ALIGN_PARAGRAPH from docx.oxml.ns import qn from docx.oxml import OxmlElement doc = Document() # --- Page margins --- from docx.shared import Inches section = doc.sections[0] section.top_margin = Inches(1) section.bottom_margin = Inches(1) section.left_margin = Inches(1.2) section.right_margin = Inches(1.2) # --- Styles --- normal_style = doc.styles['Normal'] normal_style.font.name = 'Times New Roman' normal_style.font.size = Pt(12) # Heading 1 style h1 = doc.styles['Heading 1'] h1.font.name = 'Times New Roman' h1.font.size = Pt(14) h1.font.bold = True h1.font.color.rgb = RGBColor(0x1F, 0x49, 0x7D) # Heading 2 style h2 = doc.styles['Heading 2'] h2.font.name = 'Times New Roman' h2.font.size = Pt(12) h2.font.bold = True h2.font.color.rgb = RGBColor(0x2E, 0x74, 0xB5) # --- Title Page Header --- title = doc.add_paragraph() title.alignment = WD_ALIGN_PARAGRAPH.CENTER run = title.add_run('PG ENT Model Answer Script') run.font.name = 'Times New Roman' run.font.size = Pt(16) run.font.bold = True run.font.color.rgb = RGBColor(0x1F, 0x49, 0x7D) subtitle = doc.add_paragraph() subtitle.alignment = WD_ALIGN_PARAGRAPH.CENTER r2 = subtitle.add_run('10-Mark Questions | MS ENT Examinations') r2.font.name = 'Times New Roman' r2.font.size = Pt(12) r2.font.italic = True r2.font.color.rgb = RGBColor(0x59, 0x59, 0x59) doc.add_paragraph('') # spacer # ===================================================================== # QUESTION 1: AI IN ENT # ===================================================================== doc.add_heading('Q1. Write a note on Artificial Intelligence (AI) in ENT.', level=1) marks_p = doc.add_paragraph() marks_p.alignment = WD_ALIGN_PARAGRAPH.RIGHT mr = marks_p.add_run('[10 Marks]') mr.font.bold = True mr.font.size = Pt(11) mr.font.color.rgb = RGBColor(0xC0, 0x00, 0x00) doc.add_heading('Introduction', level=2) doc.add_paragraph( 'Artificial Intelligence (AI) refers to the simulation of human intelligence processes by computer systems, ' 'encompassing machine learning (ML), deep learning (DL), and natural language processing (NLP). In otolaryngology-' 'head and neck surgery (ENT), AI applications have expanded rapidly across diagnosis, surgical planning, ' 'intraoperative assistance, and prognosis, representing a major paradigm shift in clinical practice.' ) doc.add_heading('1. AI in Audiology and Hearing', level=2) p = doc.add_paragraph(style='List Bullet') p.add_run('Automated Audiometry: ').bold = True p.add_run('AI algorithms analyse pure-tone audiograms and ABR (Auditory Brainstem Response) waveforms to classify ' 'hearing loss type and severity with accuracy comparable to audiologists.') p = doc.add_paragraph(style='List Bullet') p.add_run('Hearing Aid Fitting: ').bold = True p.add_run('Machine learning models personalise hearing aid parameters based on individual audiometric profiles and ' 'patient-reported outcomes, reducing trial-and-error fitting.') p = doc.add_paragraph(style='List Bullet') p.add_run('Newborn Hearing Screening: ').bold = True p.add_run('AI-assisted OAE (Oto-acoustic Emission) analysis reduces false-positive referral rates in universal ' 'newborn hearing screening programmes.') p = doc.add_paragraph(style='List Bullet') p.add_run('Cochlear Implant Programming: ').bold = True p.add_run('Deep learning models predict optimal mapping parameters for cochlear implant recipients, shortening ' 'rehabilitation time.') doc.add_heading('2. AI in Otology', level=2) p = doc.add_paragraph(style='List Bullet') p.add_run('Tympanic Membrane Analysis: ').bold = True p.add_run('Convolutional Neural Networks (CNNs) trained on otoscopic images can diagnose normal TM, otitis media ' '(acute, effusion, chronic), perforation, and cholesteatoma with high sensitivity and specificity.') p = doc.add_paragraph(style='List Bullet') p.add_run('Cholesteatoma Detection: ').bold = True p.add_run('AI analysis of high-resolution MRI (non-echo-planar DWI sequences) improves detection and staging ' 'of cholesteatoma pre-operatively, guiding surgical approach (CWU vs CWD mastoidectomy).') p = doc.add_paragraph(style='List Bullet') p.add_run('Surgical Navigation: ').bold = True p.add_run('Robotic cochlear implant surgery systems use AI-guided drill paths, protecting the round window, ' 'facial nerve, and ossicles with sub-millimetre precision (e.g., RobOtol system).') doc.add_heading('3. AI in Rhinology / Sinonasal Disease', level=2) p = doc.add_paragraph(style='List Bullet') p.add_run('CT/MRI Analysis: ').bold = True p.add_run('AI tools automatically segment paranasal sinuses on CT scans, calculate Lund-Mackay scores, ' 'identify anatomical variants (Haller cells, Onodi cells, Keros classification), and predict ' 'surgical difficulty in FESS (Functional Endoscopic Sinus Surgery).') p = doc.add_paragraph(style='List Bullet') p.add_run('Endoscopic Image Analysis: ').bold = True p.add_run('Real-time CNN-based systems during nasal endoscopy can identify polyps, mucosal oedema, ' 'and sinonasal tumours, alerting surgeons to areas requiring biopsy.') p = doc.add_paragraph(style='List Bullet') p.add_run('Smell and Taste Disorders: ').bold = True p.add_run('AI-based olfactory testing platforms correlate olfactometric data with aetiological diagnosis ' '(post-viral, Parkinson\'s, sinonasal).') doc.add_heading('4. AI in Head and Neck Oncology', level=2) p = doc.add_paragraph(style='List Bullet') p.add_run('Pathology: ').bold = True p.add_run('AI-powered digital pathology (whole-slide image analysis) identifies malignant transformation, ' 'tumour grade, perineural invasion, and HPV status in H&N squamous cell carcinomas with accuracy ' 'exceeding human pathologists in controlled studies.') p = doc.add_paragraph(style='List Bullet') p.add_run('Radiomics: ').bold = True p.add_run('High-dimensional feature extraction from CT/MRI/PET images (radiomics) predicts local recurrence, ' 'nodal spread, and response to chemoradiotherapy, enabling personalised treatment decisions.') p = doc.add_paragraph(style='List Bullet') p.add_run('Radiotherapy Planning: ').bold = True p.add_run('Auto-segmentation tools using deep learning delineate gross tumour volume (GTV), clinical target ' 'volume (CTV), and organs at risk (OAR) on planning CT/MRI, reducing clinician time and inter-observer ' 'variability in IMRT/IGRT planning.') p = doc.add_paragraph(style='List Bullet') p.add_run('Prognosis and Survival Prediction: ').bold = True p.add_run('ML models integrating clinical, pathological, and genomic data provide personalised survival estimates ' 'and recurrence risk scores, aiding shared decision-making.') doc.add_heading('5. AI in Laryngology / Voice Disorders', level=2) p = doc.add_paragraph(style='List Bullet') p.add_run('Voice Analysis: ').bold = True p.add_run('AI acoustic analysis of voice signals detects dysphonia, laryngeal carcinoma, vocal cord palsy, ' 'and spasmodic dysphonia from digital voice recordings, enabling remote screening.') p = doc.add_paragraph(style='List Bullet') p.add_run('Laryngoscopic Image Interpretation: ').bold = True p.add_run('CNNs classify laryngeal lesions (nodules, polyps, Reinke\'s oedema, papilloma, carcinoma) during ' 'videoendoscopy with high diagnostic accuracy, potentially reducing need for cold-cup biopsies in ' 'benign lesions.') p = doc.add_paragraph(style='List Bullet') p.add_run('Swallowing Assessment: ').bold = True p.add_run('AI analysis of FEES (Flexible Endoscopic Evaluation of Swallowing) and videofluoroscopy detects ' 'aspiration events and pharyngeal dysmotility patterns, assisting speech and language therapists.') doc.add_heading('6. AI in Thyroid and Parotid Pathology', level=2) p = doc.add_paragraph(style='List Bullet') p.add_run('Ultrasound-guided thyroid nodule assessment: ').bold = True p.add_run('AI scoring systems (e.g., ACR TIRADS augmented with ML) improve sensitivity for malignancy ' 'prediction and reduce unnecessary fine-needle aspiration cytology (FNAC).') p = doc.add_paragraph(style='List Bullet') p.add_run('Cytology: ').bold = True p.add_run('AI image analysis of FNAC slides (Bethesda classification) reduces indeterminate cytology rates ' 'and guides molecular testing decisions.') doc.add_heading('7. AI in Sleep Medicine / Snoring', level=2) p = doc.add_paragraph(style='List Bullet') p.add_run('Polysomnography Analysis: ').bold = True p.add_run('AI algorithms automate sleep-stage scoring, AHI calculation, and detection of obstructive vs ' 'central apnoeas with accuracy matching manual scoring.') p = doc.add_paragraph(style='List Bullet') p.add_run('Wearable Devices: ').bold = True p.add_run('AI-enabled home monitoring devices (oximetry, microphone-based snore analysis) screen for OSAS ' '(Obstructive Sleep Apnoea Syndrome) in primary care settings.') doc.add_heading('8. Limitations and Ethical Considerations', level=2) p = doc.add_paragraph(style='List Bullet') p.add_run('Dataset bias: ').bold = True p.add_run('AI models trained predominantly on Western or tertiary-care datasets may underperform in diverse ' 'populations.') p = doc.add_paragraph(style='List Bullet') p.add_run('Explainability (Black-box problem): ').bold = True p.add_run('Many deep learning models lack interpretability, making clinical trust and regulatory approval ' 'challenging.') p = doc.add_paragraph(style='List Bullet') p.add_run('Regulatory frameworks: ').bold = True p.add_run('AI-based medical devices require approval from bodies such as USFDA (US), CE mark (Europe), ' 'or CDSCO (India) before clinical deployment.') p = doc.add_paragraph(style='List Bullet') p.add_run('Data privacy: ').bold = True p.add_run('Handling of patient imaging and voice data must comply with HIPAA/GDPR-equivalent regulations.') doc.add_heading('Conclusion', level=2) doc.add_paragraph( 'AI is poised to become an indispensable tool in ENT practice. Its greatest current impact lies in ' 'image analysis (otoscopy, endoscopy, radiology), voice diagnostics, and oncology planning. However, ' 'AI should be viewed as augmenting rather than replacing the clinician, emphasising the continued ' 'importance of clinical judgment, patient interaction, and surgical expertise.' ) # Page break between questions doc.add_page_break() # ===================================================================== # QUESTION 2: IGRT # ===================================================================== doc.add_heading('Q2. Write a note on Image-Guided Radiotherapy (IGRT).', level=1) marks_p2 = doc.add_paragraph() marks_p2.alignment = WD_ALIGN_PARAGRAPH.RIGHT mr2 = marks_p2.add_run('[10 Marks]') mr2.font.bold = True mr2.font.size = Pt(11) mr2.font.color.rgb = RGBColor(0xC0, 0x00, 0x00) doc.add_heading('Introduction / Definition', level=2) doc.add_paragraph( 'Image-Guided Radiotherapy (IGRT) is an advanced radiotherapy technique in which imaging is performed ' 'immediately before and/or during each treatment fraction to verify patient positioning and tumour location, ' 'and to correct for any inter-fraction or intra-fraction variations. This allows more precise targeting of ' 'the tumour while reducing radiation dose to surrounding normal structures (organs at risk, OAR), thereby ' 'improving the therapeutic ratio.' ) doc.add_heading('Background and Rationale', level=2) doc.add_paragraph( 'Modern radiotherapy for head and neck cancers (and other sites) uses IMRT (Intensity-Modulated Radiotherapy) ' 'or VMAT (Volumetric Modulated Arc Therapy) to deliver highly conformal dose distributions. However, these ' 'techniques are highly sensitive to positional errors because the steep dose gradients mean even small shifts ' 'can result in tumour underdosage or excess dose to critical organs. Sources of geometric uncertainty include:' ) p = doc.add_paragraph(style='List Bullet') p.add_run('Set-up error: ').bold = True p.add_run('Day-to-day variation in patient positioning despite immobilisation devices (thermoplastic shells/masks).') p = doc.add_paragraph(style='List Bullet') p.add_run('Organ motion: ').bold = True p.add_run('Respiratory motion (lung/abdomen), cardiac pulsation, swallowing, bowel peristalsis.') p = doc.add_paragraph(style='List Bullet') p.add_run('Anatomical changes: ').bold = True p.add_run('Tumour shrinkage, weight loss, oedema resolution during prolonged treatment (e.g., 7-week head and ' 'neck chemoradiation), altering dosimetry.') doc.add_heading('Imaging Modalities Used in IGRT', level=2) doc.add_heading('(a) Cone Beam CT (CBCT)', level=2) doc.add_paragraph( 'The most widely used IGRT technique. A kilovoltage (kV) or megavoltage (MV) X-ray source mounted on ' 'the linear accelerator (linac) rotates around the patient to acquire volumetric 3D images in the treatment ' 'position. These are registered to the planning CT to detect and correct shifts in 6 degrees of freedom ' '(translational x/y/z + rotational pitch/roll/yaw). Used routinely in head and neck, prostate, lung, and ' 'abdominal cancer radiotherapy.' ) doc.add_heading('(b) Megavoltage CT (MVCT)', level=2) doc.add_paragraph( 'Used in helical tomotherapy systems (e.g., TomoTherapy). The MV beam itself is used to generate CT images ' 'immediately before treatment. Soft-tissue contrast is lower than kVCT but adequate for bony landmark ' 'alignment and gross tumour visualisation.' ) doc.add_heading('(c) 2D Orthogonal kV X-rays (Electronic Portal Imaging)', level=2) doc.add_paragraph( 'Two orthogonal planar kV images taken before treatment and registered to digitally reconstructed radiographs ' '(DRR) from the planning CT. Simpler and faster than CBCT. Adequate for bony alignment in sites where ' 'inter-fraction soft-tissue motion is limited. Implanted radio-opaque fiducial markers (gold seeds) ' 'enhance accuracy, particularly for soft-tissue targets such as the prostate or pancreas.' ) doc.add_heading('(d) MRI-Guided Radiotherapy (MRgRT)', level=2) doc.add_paragraph( 'The latest advance in IGRT, combining a linear accelerator with an MRI scanner (MR-Linac, e.g., ' 'Elekta Unity, ViewRay MRIdian). MRI provides superior soft-tissue contrast compared to CT or CBCT, ' 'enabling real-time tumour visualisation and adaptive re-planning within the treatment session. ' 'Particularly advantageous for pancreas, liver, prostate, and rectal cancers where soft-tissue definition ' 'is critical. Adaptive MRgRT (online plan adaptation based on daily MRI) represents the highest level of ' 'treatment precision currently achievable.' ) doc.add_heading('(e) Ultrasound-Guided Radiotherapy', level=2) doc.add_paragraph( 'Transabdominal or transperineal ultrasound is used to align the prostate before treatment sessions. ' 'Non-ionising, low cost, but operator-dependent and limited to selected sites.' ) doc.add_heading('(f) Surface-Guided Radiotherapy (SGRT)', level=2) doc.add_paragraph( 'Optical surface scanning using structured light or stereoscopic cameras monitors external patient ' 'surface in real time. Useful for breast radiotherapy, deep inspiration breath hold (DIBH) technique, ' 'and intracranial SRS/SBRT. Acts as a surrogate for internal target position.' ) doc.add_heading('IGRT in Head and Neck Cancer (ENT Relevance)', level=2) doc.add_paragraph( 'In the context of head and neck squamous cell carcinoma (HNSCC) treated with chemoradiation, IGRT is ' 'essential for the following reasons:' ) p = doc.add_paragraph(style='List Bullet') p.add_run('PTV Margin Reduction: ').bold = True p.add_run('Daily CBCT-based positional verification allows PTV margins to be reduced from 5-10 mm (non-IGRT) ' 'to 3-5 mm, reducing dose to the parotid glands, spinal cord, brainstem, and pharyngeal constrictors ' '(Scott-Brown\'s Otorhinolaryngology, 7th ed.).') p = doc.add_paragraph(style='List Bullet') p.add_run('Weight Loss and Anatomical Adaptation: ').bold = True p.add_run('Patients undergoing 7-week concurrent chemoradiation often lose significant weight and show tumour ' 'regression. Weekly CBCT assessment detects mask loosening and anatomical change; if significant ' 'dosimetric deviation is detected, an adaptive re-plan (re-CT and replanning) is triggered.') p = doc.add_paragraph(style='List Bullet') p.add_run('Dose Escalation: ').bold = True p.add_run('IGRT facilitates dose escalation to the GTV (e.g., simultaneous integrated boost to 70 Gy) while ' 'maintaining elective nodal dose at 56 Gy, exploiting the therapeutic window.') p = doc.add_paragraph(style='List Bullet') p.add_run('Reduced Xerostomia: ').bold = True p.add_run('Precise sparing of parotid glands (mean dose < 26 Gy), pharyngeal constrictors, and oral cavity ' 'reduces incidence of severe xerostomia, dysphagia, and trismus - the major late toxicities of ' 'H&N chemoradiation.') doc.add_heading('Adaptive Radiotherapy (ART)', level=2) doc.add_paragraph( 'Adaptive RT is the process of modifying the treatment plan during the course of radiotherapy in response ' 'to observed anatomical or biological changes detected by IGRT imaging. Two approaches exist:' ) p = doc.add_paragraph(style='List Bullet') p.add_run('Offline ART: ').bold = True p.add_run('Images from the first few fractions are reviewed, and a new plan is generated and approved before ' 'the next treatment fraction.') p = doc.add_paragraph(style='List Bullet') p.add_run('Online ART: ').bold = True p.add_run('Daily plan adaptation performed on the treatment couch (e.g., MR-Linac). The patient remains ' 'immobilised while a new plan is generated, approved, and delivered within the same session ' '(typically 15-45 minutes total).') doc.add_heading('Stereotactic Body Radiotherapy (SBRT) / SABR and IGRT', level=2) doc.add_paragraph( 'IGRT is a prerequisite for SBRT/SABR and stereotactic radiosurgery (SRS). These techniques deliver ' 'ablative doses (8-20 Gy/fraction) in 1-5 fractions to small targets with sub-millimetre accuracy. ' 'Fiducial markers, CBCT, or MRI guidance is mandatory to ensure geometric accuracy given the steep ' 'dose gradients and high fraction sizes.' ) doc.add_heading('Quality Assurance in IGRT', level=2) p = doc.add_paragraph(style='List Bullet') p.add_run('Machine QA: ').bold = True p.add_run('Daily output checks, imaging isocentre calibration, couch shift accuracy verification.') p = doc.add_paragraph(style='List Bullet') p.add_run('Dosimetric verification: ').bold = True p.add_run('Pre-treatment plan delivery verification using phantom (EPID or 3D dosimetry systems).') p = doc.add_paragraph(style='List Bullet') p.add_run('In-vivo dosimetry: ').bold = True p.add_run('EPID-based transit dosimetry or MOSFET detectors verify actual delivered dose to patient.') doc.add_heading('Advantages of IGRT', level=2) rows = [ ('Improved tumour targeting', 'Higher local control rates'), ('Reduced set-up margins', 'Less normal tissue irradiation'), ('Anatomical adaptation', 'Compensates for weight loss/tumour shrinkage'), ('Facilitates dose escalation', 'Potential survival improvement'), ('Enables SBRT/SRS', 'Ablative treatment in fewer fractions'), ] table = doc.add_table(rows=1, cols=2) table.style = 'Light Shading Accent 1' hdr = table.rows[0].cells hdr[0].text = 'Advantage' hdr[1].text = 'Clinical Benefit' for adv, ben in rows: row = table.add_row().cells row[0].text = adv row[1].text = ben doc.add_paragraph('') # spacer after table doc.add_heading('Limitations', level=2) p = doc.add_paragraph(style='List Bullet') p.add_run('Additional radiation dose from imaging (especially CBCT): ').bold = True p.add_run('Though small (1-3 cGy/fraction), cumulative imaging dose over 35 fractions is a consideration, ' 'particularly in paediatric patients.') p = doc.add_paragraph(style='List Bullet') p.add_run('Increased treatment time and resource intensity: ').bold = True p.add_run('Daily imaging adds 5-15 minutes per fraction; requires dedicated RTTs (Radiotherapy Technologists) ' 'trained in image registration.') p = doc.add_paragraph(style='List Bullet') p.add_run('Cost: ').bold = True p.add_run('MR-Linac systems are significantly more expensive than conventional linacs (~4-5 times higher ' 'capital cost), limiting availability in low/middle income settings.') p = doc.add_paragraph(style='List Bullet') p.add_run('Intra-fraction motion: ').bold = True p.add_run('CBCT captures position at the start of treatment but does not account for motion during beam ' 'delivery; MRgRT and 4D gating systems partially address this.') doc.add_heading('Conclusion', level=2) doc.add_paragraph( 'IGRT represents a pivotal advance in radiotherapy precision, particularly for head and neck cancers ' 'treated with concurrent chemoradiation. By enabling daily positional verification and adaptive replanning, ' 'IGRT reduces geometric uncertainty, allows margin reduction, and delivers higher doses to the tumour ' 'with reduced toxicity to critical structures. MR-guided radiotherapy further extends these capabilities ' 'through real-time soft-tissue visualisation, and is the current frontier of image-guided precision ' 'oncology. In ENT practice, a thorough understanding of IGRT is essential for coordinating ' 'multidisciplinary head and neck cancer care.' ) # Save out_path = '/tmp/workspace/ent-pg-answers/PG_ENT_10Mark_Answers.docx' doc.save(out_path) print(f'Saved: {out_path}') PYEOF python3 /tmp/workspace/ent-pg-answers/create_doc.py

python3 -c " from docx import Document doc = Document('/tmp/workspace/ent-pg-answers/PG_ENT_10Mark_Answers.docx') for i, p in enumerate(doc.paragraphs[:20]): print(f'[{p.style.name}] {p.text[:80]}') "