Faculty Bibliography

PURPOSE/OBJECTIVE:To characterize how the adoption of virtual residency interviews (2020-2021 cycle) has impacted the geographic distribution of radiology resident matches. METHODS:University-based interventional (IR) and diagnostic radiology (DR) residency programs from 2017 to 2021 were identified using a national residency database (FRIEDA). Public applicant data were obtained from official residency program websites. Medical schools and residency programs were categorized by US census regions. Geographic applicant distribution before and after the initiation of virtual interviews was statistically assessed using Chi-square tests. The effect of virtual interviews on the probability of matching within the same geographic region as one's medical school was evaluated with multivariate logistic regression. RESULTS:4358 radiology residents (88% diagnostic, 12% interventional) matched at 102 radiology programs during the study period. 71% (n = 3115 residents) had data available for analysis. 56.3% of DR and 49.3% of IR residents matched in the same geographic region as their medical school. The geographic distribution of applicants who matched at Southern IR residency programs significantly changed after implementation of virtual interviews (p < 0.0001). Virtual interviews did not increase the odds of matching in the same region as one's medical school for IR (OR 1.11, p = 0.08) or DR (OR 1.01, p = 0.58) applicants. Top-20 ranked DR programs had lower odds of in-region matches (OR 0.87, p < 0.001). CONCLUSION/CONCLUSIONS:With few exceptions, shifting to virtual residency interviews did not significantly affect the geographic distribution of IR or DR residency matches. Top-ranked DR programs match more regionally diverse applicants.

OBJECTIVE:Microelectrode recording (MER) is used to identify the subthalamic nucleus (STN) during deep brain stimulation (DBS) surgery. Automated STN detection typically involves extracting quantitative features from MERs for classifier training. This study evaluates the ability of feature selection to identify optimal feature combinations for automated STN localization. METHODS:We extracted 13 features from 65 MERs for classifier training. For logistic regression (LR) classification, we compared classifiers identified by feature selection to those containing all possible feature combinations. We used classification error as our metric with hold-one-patient-out cross-validation. We also compared patient-specific vs. independent normalization on classifier performance. RESULTS:Feature selection and patient-specific normalization were superior to non-optimized, patient-independent classifiers. Feature selection, patient-specific normalization, and both produced relative error reductions of 4.95%, 31.36%, and 38.92%, respectively. Three of four feature-selected LR classifiers performed better than 99% of classifiers with all possible feature combinations. Optimal feature combinations were not predictable from individual feature performance. CONCLUSIONS:Feature selection reduces classification error in automated STN localization from MERs. Additional improvement from patient-specific normalization suggests these approaches are necessary for clinically reliable automation of MER interpretation. SIGNIFICANCE/CONCLUSIONS:These findings represent an incremental advance in automated functional localization of STN from MER in DBS surgery.