Research Protocol

Diagnostic Performance of CT-like Images for Lumbar Pedicle Screw Planning and Spinal Canal Area Measurement: A Comparative Study with Conventional CT and MRI

1. BACKGROUND AND RATIONALE

2. OBJECTIVES

2.1 Primary Objective

2.2 Secondary Objectives

1. To assess agreement between CT-like images and T2-weighted MRI for spinal canal area (SCA) measurement at L1-S1.
2. To determine inter-reader and intra-reader reliability for pedicle and SCA measurements on CT-like images.
3. To identify patient or imaging factors associated with reduced agreement (e.g., obesity, metallic implants, degenerative changes, osteoporosis).
4. To evaluate the feasibility of CT-like images as a standalone tool for comprehensive preoperative lumbar spine assessment without conventional CT.

2.3 Exploratory Objective

3. STUDY DESIGN

• Conventional CT: reference standard for pedicle width and pedicle length
• T2-weighted MRI: reference standard for spinal canal area

4. STUDY POPULATION

4.1 Eligibility Criteria

• Adult patients ≥18 years of age
• Clinical indication for lumbar spine MRI (degenerative disc disease, spinal stenosis, radiculopathy, pre-surgical evaluation)
• Conventional lumbar CT also clinically indicated or performed within ≤4 weeks of MRI
• Imaging to include at least L1-L5 vertebral levels on both modalities
• Ability to provide written informed consent

• Previous lumbar spinal instrumentation (existing metallic implants causing susceptibility artifact)
• MRI contraindications (pacemakers, non-MRI-compatible implants, severe claustrophobia)
• CT or MRI of non-diagnostic quality (motion artifact, incomplete coverage of L1-S1)
• Pathological fractures, spinal tumors, or infectious spondylodiscitis distorting normal bony anatomy
• Interval between CT and MRI >4 weeks (to minimize anatomic change)
• Known spinal deformity with Cobb angle >20° (scoliosis altering pedicle orientation significantly)

4.2 Sample Size

• Expected ICC = 0.95 (based on Ogihara et al., 2026)
• Null hypothesis ICC (ICC₀) = 0.80 (minimum acceptable agreement)
• α = 0.05, power = 0.80
• Two-sided test

5. IMAGING PROTOCOLS

5.1 MRI Protocol

• Sagittal T1-weighted fast spin echo (FSE)
• Sagittal T2-weighted FSE
• Axial T2-weighted FSE at levels of interest

• 3D gradient echo or ultrashort echo time (UTE) sequence optimized for bone contrast
• Alternatively: Dixon-based 3D T1 sequence with fat-water separation for sCT reconstruction using a validated deep learning algorithm embedded in the MRI system (e.g., Siemens mDixon, Philips MR-only workflow)
• Slice thickness: ≤1.5 mm
• Field of view: covering L1 to S1
• Reconstruction: isotropic 3D volume with multiplanar reformation capability

5.2 CT Protocol

• Multi-detector CT (MDCT), minimum 64-slice
• Tube voltage: 120 kVp (adjusted per patient weight)
• Tube current: weight-adapted with automatic exposure control
• Reconstruction: 1.0-1.5 mm axial slices, bone kernel (B60 or equivalent)
• Field of view: L1-S1 inclusive
• MPR: sagittal and coronal reformations at 1.5 mm thickness

6. MEASUREMENTS

6.1 Pedicle Measurements (CT-like images vs. CT)

6.2 Spinal Canal Area (T2-weighted MRI vs. CT-like images)

7. READERS AND BLINDING

• Two independent readers (Reader 1: musculoskeletal radiologist with ≥3 years post-fellowship experience; Reader 2: spine surgeon or radiology trainee with spine imaging experience ≥2 years)
• Readers will be blinded to:
  • The modality being reviewed (CT vs. CT-like images presented in random sequence)
  • The other reader's measurements
  • Clinical outcome data
• A second reading session will be performed by Reader 1 ≥4 weeks after the first to assess intra-reader reliability.
• Readers will be trained on a practice dataset of 5 cases before study measurements begin.

8. STATISTICAL ANALYSIS

8.1 Primary Analysis - Agreement

• Intraclass Correlation Coefficient (ICC): Two-way mixed effects model, absolute agreement type (ICC(2,1)) for each measurement (pedicle width, pedicle length, SCA), separately for each lumbar level and pooled across levels.
• ICC interpretation: <0.50 poor, 0.50-0.74 moderate, 0.75-0.89 good, ≥0.90 excellent (Koo & Mae, 2016).
• 95% confidence intervals reported for all ICC values.

8.2 Bland-Altman Analysis

• Calculation of mean difference (bias) and 95% limits of agreement (LOA = mean difference ± 1.96 SD) for each measurement pair.
• Proportional bias assessed by regression of difference vs. mean.
• Clinically acceptable limits pre-specified as: pedicle width difference ≤1 mm, pedicle length difference ≤2 mm, SCA difference ≤10%.

8.3 Secondary Analyses

• Inter-reader reliability: ICC between Reader 1 and Reader 2 on the same modality.
• Intra-reader reliability: ICC between Reader 1 session 1 and session 2.
• Subgroup analyses: Agreement stratified by vertebral level (L1-L5), patient BMI (≤25, 25-30, >30 kg/m²), and degree of degenerative change.
• Equivalence testing: One-sided 95% confidence intervals to test whether mean differences are within pre-specified equivalence margins.

8.4 Software

9. ETHICAL CONSIDERATIONS

• Protocol submission to the Institutional Review Board (IRB)/Ethics Committee prior to any recruitment.
• All participants will provide written informed consent before enrollment.
• CT scans will only be included from those ordered as part of standard clinical care; no additional radiation exposure will be incurred by study participation.
• Patient data de-identified prior to analysis; stored on a secure, password-protected server compliant with institutional data governance policies.
• The study will be conducted in accordance with the Declaration of Helsinki (2013) and applicable national research regulations.
• Registered prospectively on a recognized clinical trial registry (e.g., ClinicalTrials.gov or UMIN) prior to first enrollment.

10. STUDY LIMITATIONS (TO BE ACKNOWLEDGED)

1. Retrospective component / mixed design: Including retrospectively collected paired datasets may introduce selection bias.
2. Single center: Results may not generalize to different MRI hardware or sCT algorithms.
3. CT-like algorithm variability: Multiple vendors use different sCT generation pipelines; findings may differ between systems.
4. Gold standard for SCA: T2-weighted MRI is used as reference, but MRI SCA measurement itself has known variability in severe stenosis where the CSF signal is obliterated.
5. No direct surgical outcome correlation unless the exploratory objective is achievable within the study timeframe.

11. TIMELINE

12. REFERENCES

1. Ogihara A, Fukuda T, Katsumi S, Ojiri H. Diagnostic Performance of CT-like Images for Lumbar Pedicle Screw Planning and Spinal Canal Area Measurement: A Comparative Study with Conventional CT and MRI. Tomography. 2026; doi:10.3390/tomography12030040 [PMID: 41893835]
2. Sankar S, McDonnell JM, Darwish S, Butler JS. The potential role of synthetic computed tomography in spinal surgery: generation, applications, and implications for future clinical practice. Asian Spine J. 2024; doi:10.31616/asj.2024.0197 [PMID: 39763361]
3. Altorfer FCS et al. Deep-learning reconstructed lumbar spine 3D MRI for surgical planning: pedicle screw placement and geometric measurements compared to CT. Eur Spine J. 2024. [PMID: 38472429]
4. Scientific Reports (2025). Evaluation of MRI-based synthetic CT for lumbar degenerative disease: a comparison with CT. doi:10.1038/s41598-025-05399-x
5. J Clin Med (2025). Generating a Synthetic Lumbar CT from a Standard MRI for Surgical Planning.
6. Koo TK, Mae AY. A Guideline of Selecting and Reporting Intraclass Correlation Coefficients for Reliability Research. J Chiropr Med. 2016;15(2):155-163.
7. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1(8476):307-310.

APPENDIX A: Data Collection Form (Template)