MétaCan
Menu
Back to cohort

S762 Evaluation of Clinical Variables, Radiological Visual Analog Scoring, and Radiomics Features on MR Enterography for Characterizing Severe Inflammation and Fibrosis in Stricturing Crohn’s Disease

2022· article· en· W4320064976 on OpenAlexaboutno aff
Joseph Sleiman, Prathyush Chirra, Namita Gandhi, Ilyssa O. Gordon, Satish E. Viswanath, Florian Rieder

Bibliographic record

VenueThe American Journal of Gastroenterology · 2022
Typearticle
Languageen
FieldMedicine
TopicRadiomics and Machine Learning in Medical Imaging
Canadian institutionsnot available
Fundersnot available
KeywordsMedicineFibrosisRadiologyStenosisRadiomicsHistopathologyInflammationPathologyInternal medicine

Abstract

fetched live from OpenAlex

Introduction: Current non-invasive cross-sectional imaging modalities such as MR enterography (MRE) offer excellent diagnostic accuracy of Crohn’s disease (CD) strictures, but cannot accurately determine the extent of stricture fibrosis and inflammation. Radiomics, a quantitative image extraction analysis technology, may offer a solution. We present initial results for a machine-reader evaluation of severe inflammation and fibrosis in CD strictures via quantitative radiomic features and expert radiologist scoring of MRE. Methods: In this retrospective, single center, IRB-approved study, 51 patients (n=34 for discovery; n=17 for hold-out validation) had confirmed stricturing CD on MRE and histopathology from surgery within 15 weeks of MRE. Histopathological Stenosis Therapy & Research (STAR) scoring of specimens (range 0-100, scores ≥50 =severe) was the reference standard for both inflammation and fibrosis. An expert radiologist coordinated with the scoring pathologist to annotate the resected strictures on MRE and provide a global visual analog score (VAS, 0-100) assessment of inflammation and chronic non-inflammatory findings (fibrosis). 1852 3D radiomic features were extracted from the stricture regions on MRE, from which the most relevant feature subsets were identified via cross-validated machine learning analysis in the discovery cohort for differentiating between severe vs less severe inflammation and fibrosis. Radiomic features and VAS scores were evaluated against pathology-defined severe inflammation and fibrosis in the validation cohort via ROC analysis. Results: Two distinct sets of radiomic features capturing textural heterogeneity (patterns, local entropy) within strictures were significantly associated (p< 0.01) with severe inflammation and severe fibrosis; across both discovery (AUC=0.66, 0.76) and hold-out validation (AUCs =0.71,0.83) (Figure). Radiological VAS had an AUC=0.68 for identifying severe inflammation and AUC =0.47 for severe fibrosis. Combining radiomic features and VAS had no significant impact on predictor performance. Clinical variables including sex, age, Montreal classification and stricture type were not significantly associated severe inflammation or fibrosis, across discovery and validation groups (Table). Conclusion: Radiomic analysis shows improved performance in identifying severe inflammation and severe fibrosis in CD strictures on MRE compared to radiological visual assessment scoring and clinical variables.Figure 1.: Top-ranked radiomics features are distinctively associated with severe inflammation (top row, pattern-based) and severe fibrosis (bottom row, wavelets) on MRE. Also shown are radiological VAS for severe inflammation and severe fibrosis. Table 1. - Demographics and baseline clinical features of the cohort, segregating discovery and hold-out validation radiomic cohorts MRE Overall (N=51) Fibrosis Discovery Group (N=34) Fibrosis Validation Group (N=17) Inflammation Discovery Group (N=34) Inflammation Validation Group (N=17) Factor N Statistics N Statistics N Statistics P-value N Statistics N Statistics P-value Male Sex, n (%) 51 26 (51) 34 16 (47) 17 10 (57) 0.43 a 34 15 (44) 17 11 (65) 0.17 a Diagnosis age of IBD, median (range), yrs 51 21 (4-90) 34 24 (10-90) 17 20 (2-62) 0.88 c 34 20.5 (5-67) 17 25 (4-90) 0.79 c Diagnosis age of Stricture, median (range), yrs 51 32 (11-90) 34 30.5 (19-90) 17 33 (11-69) 0.78 c 34 29.5 (11-71) 17 35 (20-90) 0.24 c Age at MRE, median (range), yrs 51 34 (18-91) 34 33 (19-91) 17 36 (18-69) 0.83 c 34 31 (18-71) 17 37 (22-91) 0.21 c Duration between IBD/stricture dx, median (range), years 51 8 (0-30) 34 6.5 (0-30) 17 10 (0-26) 0.62 c 34 7.5 (0-30) 17 8 (0-21) 0.8 c Duration between Stricture dx/Surgery, median (range), months 51 9 (0-145) 51 10 (0-145) 17 6 (0-121) 0.82 c 34 5.5 (0-145) 17 19 (0-78) 0.24 c Duration between MRE and resection, median (range), weeks 51 7.1 (0-15) 34 7.35 (0-13) 17 7 (0.9-15) 0.36 c 34 7.9 (0.1-15) 17 7.1 (0-14.9) 0.93 c Obstructive Symptoms at time of imaging, n (%) 51 42 (82) 34 28 (82) 17 14 (82) 1 b 34 27 (79) 17 15 (88) 0.7 b CD Montreal Classification, n (%) 51 34 17 0.64 a 34 17 0.14 a B2 (Stricturing) 23 (45) 14 (41) 9 (53) 17 (50) 6 (35) B2p (Stricturing with perianal disease) 15 (29) 11 (32) 4 (23) 11 (32) 4 (23) B3 (Fistulizing) 6 (12) 5 (15) 1 (6) 4 (12) 2 (12) B3p (Fistulizing with perianal disease) 7 (14) 4 (12) 3 (18) 2 (6) 5 (29) History of extraintestinal manifestations, n (%) 51 32 (63) 34 22 (65) 17 10 (59) 0.68 a 34 22 (65) 17 10 (59) 0.68 a Ileocecal resection prior to current stricture, n (%) 51 25 (49) 34 17 (50) 17 8 (47) 0.84 a 34 16 (47) 17 9 (53) 0.69 a Number of resections, median (range) 25 2 (1-5) 16 2 (1-5) 8 2 (1-4) 0.88 c 16 2 (1-4) 2 (1-5) 0.94 c Type of stricture, n (%) 51 34 17 1 a 34 17 1 a Naïve 27 (53) 18 (53) 9 (53) 18 (53) 9 (53) Anastomotic 24 (47) 16 (47) 8 (47) 16 (47) 8 (47) Medications for IBD < 8 weeks from imaging, n (%) 51 34 17 34 17 5-aminosalicylic-acid, oral or rectal 10 (20) 8 (24) 2 (12) 0.46 b 6 (18) 4 (24) 0.71 b Steroid, systematic 19 (37) 11 (32) 8 (47) 0.31 a 12 (35) 7 (41) 0.68 a Steroid, rectal or Budesonide 12 (24) 9 (27) 3 (18) 0.73 b 8 (24) 4 (24) 1 b Mercaptopurine or Azathioprine 12 (24) 7 (21) 5 (29) 0.5 b 10 (29) 2 (12) 0.29 b Methotrexate 2 (4) 2 (6) 0 (0) 0.55 b 1 (3) 1 (6) 1 b Certolizumab 2 (4) 2 (6) 0 (0) 0.55 b 1 (3) 1 (6) 1 b Adalimumab 13 (25) 10 (29) 3 (18) 0.5 b 8 (24) 5 (29) 0.74 b Infliximab 5 (10) 3 (9) 2 (12) 1 b 4 (12) 1 (6) 0.65 b Vedolizumab 5 (10) 4 (12) 1 (6) 0.65 b 2 (6) 3 (18) 0.32 b None 6 (12) 4 (12) 1 (6) 0.65 b 4 (12) 1 (6) 0.65 b Global Assessments by Radiologist Global Stricture Severity, median (range), 0-100 51 60 (20-100) 34 50 (20-100) 17 60 (20-100) 0.4 c 34 60 (20-100) 17 40 (20-95) 0.44 c Global Inflammation Severity, median (range), 0-100 51 40 (15-95) 34 40 (15-85) 17 50 (15-95) 0.27 c 34 40 (15-85) 17 40 (15-95) 0.7 c Global Chronic non-inflammatory changes Severity, median (range), 0-100 51 30 (5-80) 34 30 (5-80) 17 40 (5-80) 0.16 c 34 30 (5-80) 17 30 (5-70) 0.89 c Global Assessments by Pathologist Severity of inflammation, median (range), 0-100 51 66 (2-100) 34 67 (2-100) 17 64 (10-94) 0.52 c 34 61.5 (2-100) 17 67 (10-100) 0.73 c Severity of fibrosis, median (range), 0-100 51 60 (5-94) 34 60 (5-94) 17 55 (10-88) 0.93 c 34 57.5 (5-94) 17 63 (10-90) 0.93 c aChi-Square testbFisher exact testcMann Whitney U test.MRE: magnetic resonance enterography; N: Number; IBD: inflammatory bowel disease; dx: diagnosis; CD: Crohn’s disease.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

How this classification was reachedexpand

Full frame machine prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.

metaresearch head score (Codex)0.003
metaresearch head score (Gemma)0.003
Version: metacan-v3-hybrid-931329e0061cValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Observational · Consensus signal: Observational
GenreCandidate signal: Empirical · Consensus signal: Empirical
Teacher disagreement score0.003
Threshold uncertainty score0.018

Distilled classifier scores by category (both heads)

CategoryCodexGemma
Metaresearch0.0030.003
Meta-epidemiology (narrow)0.0000.000
Meta-epidemiology (broad)0.0010.001
Bibliometrics0.0010.001
Science and technology studies0.0000.000
Scholarly communication0.0010.000
Open science0.0000.000
Research integrity0.0000.000
Insufficient payload (model declined to judge)0.0030.001

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.022
GPT teacher head0.339
Teacher spread0.317 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.

The models applied no category: nothing in the taxonomy fit this work.
Study designObservational
Domainnot available
GenreEmpirical

How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".

Quick stats

Citations0
Published2022
Admission routes1
Has abstractyes

Explore more

Same venueThe American Journal of GastroenterologySame topicRadiomics and Machine Learning in Medical ImagingFrench-language works237,207