Citations

Deep learning offers significant advancements in the accuracy of prostate identification and classification, underscoring its potential for clinical integration. However, the opacity of deep learning models presents interpretability challenges, critical for their acceptance and utility in medical diagnosis and detection. While explanation methods have been proposed to demystify these models, enhancing their clinical viability, the efficacy and acceptance of these methods in medical tasks are not well documented. This pilot study investigates the effectiveness of deep learning explanation methods in clinical settings and identifies the attributes that radiologists consider crucial for explainability, aiming to direct future enhancements. This study reveals that while explanation methods can improve clinical task performance by up to 20%, their perceived usefulness varies, with some methods being rated poorly. Radiologists prefer explanation methods that are robust against noise, precise, and consistent. These preferences underscore the need for refining explanation methods to align with clinical expectations, emphasizing clarity, accuracy, and reliability. The findings highlight the importance of developing explanation methods that not only improve performance but also are tailored to meet the stringent requirements of clinical practice, thereby facilitating deeper trust and a broader acceptance of deep learning in medical diagnostics.

Objective Takayasu arteritis is a rare large vessel vasculitis with incompletely understood etiology. This study was undertaken to perform the first unbiased genome‐wide association analysis of Takayasu arteritis. Methods Two independent cohorts of patients with Takayasu arteritis from Turkey and North America were included in our study. The Turkish cohort consisted of 559 patients and 489 controls, and the North American cohort consisted of 134 patients and 1,047 controls of European ancestry. Genotyping was performed using the Omni1‐Quad and Omni2.5 genotyping arrays. Genotyping data were subjected to rigorous quality control measures and subsequently analyzed to discover genetic susceptibility loci for Takayasu arteritis. Results We identified genetic susceptibility loci for Takayasu arteritis with a genome‐wide level of significance in IL6 (rs2069837) (odds ratio [OR] 2.07, P = 6.70 × 10 −9 ), RPS9 / LILRB3 (rs11666543) (OR 1.65, P = 2.34 × 10 −8 ), and an intergenic locus on chromosome 21q22 (rs2836878) (OR 1.79, P = 3.62 × 10 −10 ). The genetic susceptibility locus in RPS9 / LILRB3 lies within the leukocyte receptor complex gene cluster on chromosome 19q13.4, and the disease risk variant in this locus correlates with reduced expression of multiple genes including the inhibitory leukocyte immunoglobulin‐like receptor gene LILRB3 ( P = 2.29 × 10 −8 ). In addition, we identified candidate susceptibility genes with suggestive levels of association ( P < 1 × 10 −5 ) with Takayasu arteritis, including PCSK5, LILRA3, PPM1G/NRBP1 , and PTK2B . Conclusion Our findings indicate novel genetic susceptibility loci for Takayasu arteritis and uncover potentially important aspects of the pathophysiology of this form of vasculitis.