Faculty Bibliography

PURPOSE/OBJECTIVE:To develop and evaluate a hybrid large language model (LLM)- and rule-based algorithm for tracking pancreatic cystic lesions (PCLs) across serial radiology reports and apply it to a single-center cohort to characterize cyst growth using first-to-last and consecutive comparisons. METHODS:This retrospective study included all abdominal imaging reports interpreted by abdominal radiologists between 2005 and 2024. A validated LLM identified reports describing PCLs. A hybrid algorithm combining LLM-extracted cyst features (size, location) with rule-based logic matched dominant cysts across serial reports. Two blinded radiologists independently annotated cyst matches in a subset of reports. Agreement between algorithm and radiologists was assessed using percent agreement and Cohen's κ. Discrepancies were adjudicated by a third radiologist for error analysis. Absolute size-based growth was assessed according to American College of Radiology (ACR) criteria, whereas cyst growth rate as a worrisome feature was assessed per Kyoto criterion, using both first-to-last and consecutive studies. Predictors of meeting the Kyoto cyst growth rate criterion were analyzed using multivariable Cox regression. RESULTS:In 58 patients with 418 longitudinal reports, agreement with the adjudicated reference standard was 96%. Cohen's κ was 0.81 (95% CI: 0.66-0.91). The overall cohort included 8,176 patients (61.8% female; mean age 67.6 ± 11.7 years) with 8,389 cysts across 28,489 reports over 40.6 ± 28.5 months (annual growth rate = 0.35 mm/year). ACR absolute growth criteria classified a higher proportion of cysts than the Kyoto cyst growth rate criterion using first-to-last examinations (16.3% vs. 7.4%, p < 0.001), whereas Kyoto cyst growth rate criterion identified a higher proportion using consecutive examinations (21.4% vs. 14%, p < 0.001). Larger baseline cyst size predicted meeting the Kyoto cyst growth rate criterion (HR 1.9-4.0; highest HR 4.0, 95% CI 3.08-5.21; p < 0.001). CONCLUSION/CONCLUSIONS:A hybrid LLM-rule-based algorithm enables reliable automated longitudinal tracking of PCLs across serial radiology reports. When applied at scale, cyst growth classification varied according to the guideline framework and interval comparison method used, and baseline cyst size was the strongest predictor of meeting Kyoto growth rate threshold.

RATIONALE AND OBJECTIVES/OBJECTIVE:Prostate multiparametric magnetic resonance imaging (MRI) is recommended for prostate cancer detection, staging, and surveillance. Incidental bladder lesions are encountered on these studies but remain under-characterized in the literature. The patient characteristics associated with malignancy for these lesions are not well defined. We evaluated the prevalence, histopathologic outcomes, clinical characteristics, and associations with malignancy for incidental bladder lesions on prostate MRI. MATERIALS AND METHODS/METHODS:A retrospective review included 31,241 patients undergoing prostate MRI examinations from January 2013 to January 2023. Imaging reports and medical records were analyzed for incidental bladder lesions, demographic data, clinical symptoms, urinalysis findings, and histopathologic outcomes. Lesions were categorized based on biopsy results or negative clinical follow-up for bladder tumors in chart review. Logistic regression analysis and receiver operating characteristic analyses were performed. RESULTS:Incidental bladder lesions occurred in 0.74% (230/31,241) of examinations, with biopsy-confirmed bladder cancer in 0.11% of patients (34/31,241) or 14.8% (34/230) of cases with lesions. In multivariable analysis, gross hematuria had the strongest association with biopsy-proven bladder cancer (OR 9.26, 95% CI 4.12-20.79, p<0.001). A logistic regression model incorporating age, smoking status, and gross hematuria yielded area under the curve of 0.762 for bladder cancer. CONCLUSION/CONCLUSIONS:Incidental bladder lesions on prostate MRI may represent opportunities for early detection of bladder cancer, but also have potential for harms related to unnecessary procedures. Considering the presence of gross hematuria, possibly stated as part of the MRI referral or patient questionnaire, could improve risk stratification of encountered bladder lesions and early cancer detection.

The ACR Incidental Findings Committee presents recommendations for managing incidental pineal cysts on CT of the head or MRI of the brain. The Pineal Cyst Subcommittee is composed of neuroradiologists and a neurosurgeon who developed the algorithms presented. These recommendations represent a combination of current published evidence as well as expert experience and opinion and were finalized by a formal consensus-building process. The recommendations address commonly encountered incidental findings in the pineal gland and are not intended to be a comprehensive review of all pineal incidental findings. The goal is to improve the quality of care by providing guidance on management of incidentally detected pineal cysts.

PURPOSE/OBJECTIVE:To evaluate the association of patient characteristics, community-level social determinants of health, and cyst risk categories with completion of follow-up recommendations for incidental Pancreatic Cystic Lesions (PCLs). METHODS:We retrospectively identified consecutive patients (2013-2023) whose MRI radiology reports described PCLs. A fine-tuned LLaMA-3.1 8B Instruct large language model was used to extract PCL features. Lesions were classified using the 2017 ACR white paper: Category 1 (low risk), Category 2 (worrisome features), or Category 3 (high-risk stigmata). We recorded demographics and follow-up imaging or endoscopic ultrasound dates. Community-level factors were characterized by the 2020 CDC Social Vulnerability Index (SVI), stratified into quartiles. The primary outcome, "inappropriate follow-up," combined late and no follow-up. Multivariable binomial regression was applied to evaluate associations with inappropriate follow-up. RESULTS:In 7,745 patients (mean age 66.3 years; 4,796 women), 92.9% (7,198/7,745) of cysts were Category 1, 6.4% (498/7,745) were Category 2, and 0.6% (49/7,745) were Category 3. Only 36.3% of patients completed appropriate follow-up, 12.1% were late, and 51.6% were lost to follow-up. Inappropriate follow-up was high in every cyst category: 64.2% in Category 1, 59.4% in Category 2 and 49.0% in Category 3. In multivariable analysis, non-English primary language (RR 1.08; 95% CI, 1.02-1.14) and residing in more vulnerable communities of the 3rd quartiles of the socioeconomic Social Vulnerability Index subcategory (RR 1.07; 95% CI, 1.02-1.12) were associated with inappropriate follow-up. Higher age-adjusted Charlson Comorbidity Index (CCI ≥ 4) (RR .84; 95% CI, .79-.88), CCI 2-3 (RR .84; 95% CI, .79-.88), and higher-risk cysts in patients under 65 years of age (RR .76; 95% CI, .65-.89) were associated with completed follow-up. CONCLUSION/CONCLUSIONS:Follow-up completion for incidental PCLs was low. Factors most consistently associated with follow-up completion were language barriers, residence in socioeconomically vulnerable communities, age-adjusted CCI and higher-risk features among those under 65 years.

Artificial intelligence (AI) applications for radiology workflow have the potential to improve patient and health-system-level outcomes through more efficient and accurate diagnosis and clinical decision making. For a variety of time-intensive steps, numerous types of applications are now available with variable reported measures and degrees of success. The tools we highlight aim to accelerate imaging acquisition, reduce cognitive and manual burden on radiologists and others involved in the care pathway, improve diagnostic accuracy, and shorten the time to clinical action based on imaging results. Most existing studies have focused on intermediate outcomes, such as task duration or time to the next step in care. In this article, we present an examination of AI applications across the medical imaging exam workflow, review examples of real-world evidence on these tools, and summarize the relevant performance metrics by application type. Beyond the more immediately acquired measures, to demonstrate benefit to patient health and economic outcomes, a more integrated assessment is necessary, and in an iterative fashion. To evolve beyond early workflow gains, interoperable tools must be tied to measurable downstream impacts, such as reduced disease severity, lower mortality, and shorter hospital stays, while we acknowledge that current empirical evaluations are limited.

BACKGROUND:Inferring risk for malignant transformation (MT) in patients with lesions diagnosed as mild or moderate oral epithelial dysplasia (low-grade OED) remains challenging. We developed two models assessing the risk of progression to high-grade OED (severe dysplasia or carcinoma in situ) or OSCC in patients with low-grade OED lesions. METHODS:We included demographic, risk habit and clinical data from participants with low-grade OED lesions enrolled in the BC Oral Cancer Prevention Program's Oral Cancer Prediction Longitudinal study. Cox proportional hazard models were fit to estimate the effects of anatomic site and toluidine blue findings and adjusted for confounders, as both are associated with MT in the literature but without a North American-specific cohort analysis. Our primary model included both variables of interest. A secondary model included only anatomic site since toluidine blue is not in widespread use. RESULTS:Five hundred and thirty-four participants with 605 lesions met final inclusion criteria, with 339 mild and 266 moderate OED at baseline. In the primary model, lesions at a high-risk anatomic site or with positive toluidine blue staining were associated with a 2.6 and 2.4-fold increased risk of progression, respectively. In the second model that did not incorporate toluidine blue, high-risk anatomic site remained a highly associated risk factor (2.7-fold increased risk of progression). CONCLUSION/CONCLUSIONS:Lesion anatomic site is associated with higher risk of MT for the general practitioner, while a specialist with access to toluidine blue results can assume additional risk associated with positive staining. These models may inform decisions for surveillance and intervention for OED.

We conducted an umbrella review to synthesize the evidence on the diagnostic performance of liquid biopsy tests for detection of head and neck squamous cell carcinoma (HNSCC). Systematic reviews (SRs) were searched in Medline, Embase, and Google Scholar through December 6, 2023. The Joanna Briggs Institute Critical Appraisal Tool for Systematic Reviews was used to assess methodological quality. Two independent reviewers extracted data. We examined the pooled sensitivity and specificity of biomarker classes. We also statistically pooled sensitivity and specificity of individual biomarkers for oral SCC in cases where meta-analysis was not yet published, since most HNSCC occurs in the oral cavity. Performance was also assessed by specimen type (saliva, serum, plasma, and whole blood). Thirty-one SRs met inclusion criteria and 21 included meta-analyses on transcriptomic, proteomic, genomic, or metabolomic biomarkers. Overall methodologic quality was moderate to high. Primary study overlap was ≥ 15 % in 9.3 % of SR pairwise comparisons. MicroRNA (miRNA) was the biomarker class represented in the most SRs (n = 19) and individual studies (n = 106). Among these, the highest sensitivity was 77 % (95 % CI, 68-84 %), observed in miRNA-21. Hypermethylated DNA was the biomarker class with the highest pooled sensitivity (86 %; 95 % CI, 60-96 %) and specificity (92 %; 95 % CI, 80-97 %) overall, and with superior performance when used in panels compared to individual markers. In studies focused on OSCC detection, no other biomarker class or fluid type demonstrated superior performance over others. In future clinical studies, panels including hypermethylated DNA merit more rigorous evaluation to establish high specificity in addition to sufficient sensitivity.