Kam, Anthony

The translation of AI-generated brain metastases (BM) segmentation into clinical practice relies heavily on diverse, high-quality annotated medical imaging datasets. The BraTS-METS 2023 challenge has gained momentum for testing and benchmarking algorithms using rigorously annotated internationally compiled real-world datasets. This study presents the results of the segmentation challenge and characterizes the challenging cases that impacted the performance of the winning algorithms. Untreated brain metastases on standard anatomic MRI sequences (T1, T2, FLAIR, T1PG) from eight contributed international datasets were annotated in stepwise method: published UNET algorithms, student, neuroradiologist, final approver neuroradiologist. Segmentations were ranked based on lesion-wise Dice and Hausdorff distance (HD95) scores. False positives (FP) and false negatives (FN) were rigorously penalized, receiving a score of 0 for Dice and a fixed penalty of 374 for HD95. The mean scores for the teams were calculated. Eight datasets comprising 1303 studies were annotated, with 402 studies (3076 lesions) released on Synapse as publicly available datasets to challenge competitors. Additionally, 31 studies (139 lesions) were held out for validation, and 59 studies (218 lesions) were used for testing. Segmentation accuracy was measured as rank across subjects, with the winning team achieving a LesionWise mean score of 7.9. The Dice score for the winning team was 0.65 ± 0.25. Common errors among the leading teams included false negatives for small lesions and misregistration of masks in space. The Dice scores and lesion detection rates of all algorithms diminished with decreasing tumor size, particularly for tumors smaller than 100 mm3. In conclusion, algorithms for BM segmentation require further refinement to balance high sensitivity in lesion detection with the minimization of false positives and negatives. The BraTS-METS 2023 challenge successfully curated well- annotated, diverse datasets and identified common errors, facilitating the translation of BM segmentation across varied environments and providing the tools for future development of personalized volumetric reports to patients undergoing BM treatment.

Background and importance Neurointervention is a very competitive specialty in the United States due to the limited number of training spots and the larger pool of applicants. The training standards are continuously updated to ensure solid training experiences. Factors affecting candidate(s) selection have not been fully established yet. Our study aims to investigate the factors influencing the selection process. Methods A 52-question survey was distributed to 93 program directors (PDs). The survey consisted of six categories: (a) Program characteristics, (b) Candidate demographics, (c) Educational credentials, (d) Personal traits, (e) Research and extracurricular activities, and (f) Overall final set of characteristics. The response rate was 59.1%. As per the programs’ characteristics, neurosurgery was the most involved specialty in running the training programs (69%). Regarding demographics, the need for visa sponsorship held the greatest prominence with a mean score of 5.9 [standard deviation (SD) 2.9]. For the educational credentials, being a graduate from a neurosurgical residency and the institution where the candidate's residency training is/was scored the highest [5.4 (SD = 2.9), 5.4 (SD = 2.5), respectively]. Regarding the personal traits, assessment by faculty members achieved the highest score [8.9 (SD = 1)]. In terms of research/extracurricular activities, fluency in English had the highest score [7.2 (SD = 1.9)] followed by peer-reviewed/PubMed-indexed publications [6.4 (SD = 2.2)]. Conclusion Our survey investigated the factors influencing the final decision when choosing the future neurointerventional trainee, including demographic, educational, research, and extracurricular activities, which might serve as valuable guidance for both applicants and programs to refine the selection process.