Quantifying and leveraging predictive uncertainty for medical image assessment

The interpretation of medical images is a challenging task, often complicated by the presence of artifacts, occlusions, limited contrast and more. Most notable is the case of chest radiography, where there is a high inter-rater variability in the detection and classification of abnormalities. This is largely due to inconclusive evidence in the data or subjective definitions of disease appearance. An additional example is the classification of anatomical views based on 2D Ultrasound images. Often, the anatomical context captured in a frame is not sufficient to recognize the underlying anatomy. Current machine learning solutions for these problems are typically limited to providing probabilistic predictions, relying on the capacity of underlying models to adapt to limited information and the high degree of label noise. In practice, however, this leads to overconfident systems with poor generalization on unseen data. To account for this, we propose a system that learns not only the probabilistic estimate for classification, but also an explicit uncertainty measure which captures the confidence of the system in the predicted output. We argue that this approach is essential to account for the inherent ambiguity characteristic of medical images from different radiologic exams including computed radiography, ultrasonography and magnetic resonance imaging. In our experiments we demonstrate that sample rejection based on the predicted uncertainty can significantly improve the ROC-AUC for various tasks, e.g., by 8% to 0.91 with an expected rejection rate of under 25% for the classification of different abnormalities in chest radiographs. In addition, we show that using uncertainty-driven bootstrapping to filter the training data, one can achieve a significant increase in robustness and accuracy. Finally, we present a multi-reader study showing that the predictive uncertainty is indicative of reader errors.     

The feasibility,effectiveness and safety of totally thoracoscopic resection for right atrial myxoma

Background Myxoma is the most frequent cardiac benign tumor which requires surgical removal even though in right atrium. Minimally invasive surgery has become the standard approach for removal of myxoma in our center. Herein, we report our initial experience of cardiac myxoma removal through mini-port totally thoracoscopic particular with focus on its feasibility, efficacy and short-term outcomes, as compared with median sternotomy. Methods From April 2015 to March 2019, 31 consecutive cases were ultimately identified and divided into two groups(Mini-port thoracoscopic group, n=14; versus conventional median sternotomic group, n=17). An echocardiography was routinely performed prior to discharge, at 1 st month, 6 th month, 12 th month and every year postoperatively. Retrospective analyses of perioperative data, postoperative outcomes and complications,and follow up data were performed. Results All the 31 patients successfully underwent right atrial myxoma removal surgery with no perioperative death, re-exploration for bleeding, residual mass or central nervous event in both groups. Intraoperative data from two group showed absence of significant difference except the intraoperative blood loss(150±50.01 m L in mini-port thoracoscopic group vs. 255.88±93.35 m L in conventional median sternomtomic group, P=0.01). There was no significant difference in perioperative blood transfusion and postoperative outcomes between the two groups. But the mini-port thoracoscopic group had a much less 24-hour postoperative drainage(136.71±148.20 m L vs. 341.17±211.29 m L, P=0.03). No significance difference could be identified between two groups about the incidence of postoperative morbidities and follow-up adverse events. Conclusion Totally mini-port video-assisted thoracoscopic approach for right atrial myxoma resection on peripheral cardiopulmonary bypass is feasible, effective with favorable safety in experienced cardiac center.[S Chin J Cardiol 2019;20(4):236-244]     

Deep Phenotyping and Prediction of Long-term Cardiovascular Disease: Optimized by Machine Learning

BackgroundPrediction of cardiovascular disease (CVD) is important in clinical practice. Machine learning (ML) may offer an improved alternative to current CVD risk stratification in individual patients. We aim to identify important predictors and compare ML models with traditional models according to their prediction performance in a large long-term follow-up cohort.MethodsThe Atherosclerosis Risk in Communities (ARIC) study was designed to study the progression of subclinical disease to cardiovascular events over a 25-year follow-up period. All phenotypic variables at visit 1 were obtained. All-cause death, CVD, and coronary heart disease were the outcomes for analysis. The ML framework involved variable selection using the random survival forest (RSF) method, model building, and 5-fold cross-validation. Model performance was evaluated by discrimination using the Harrell concordance index (C-index), accuracy using the Brier score (BS), and interpretability using the number of variables in the model.ResultsOf the 14,842 participants in ARIC, the average age was 54.2 years, with 45.2% male and 26.2% Black participants. Thirty-eight unique variables were selected in the RSF top 20 importance ranking of all 6 outcomes. Aging, hypertension, glucose metabolism, renal function, coagulation, adiposity, and sodium retention dominated the predictions of all outcomes. The ML models outperformed the regression models and established risk scores with a higher C-index, lower BS, and varied interpretability.ConclusionsThe ML framework is useful for identifying important predictors of CVD and for developing models with robust performance compared with existing risk models.     

以肺动脉高压为首发表现的先天性中枢性低通气综合征临床特征和基因分析

目的探讨先天性中枢性低通气综合征(CCHS)的临床和基因变异特征。方法分析1例首发表现为不明原因肺动脉高压的CCHS患儿的临床资料,并总结国内外文献中CCHS病例的临床特点、致病机制和基因变异情况。结果11月龄女婴,主要表现为浮肿、尿少、低血压、嗜睡、发绀、抽搐及颅内压增高。B型脑利钠肽、丙氨酸氨基转移酶升高,凝血酶原时间延长。颅脑磁共振示右侧额叶出血;超声心动图示中重度肺动脉高压。靶向捕获二代测序未发现可能的致病基因。采用Sanger法验证示患儿PHOX 2 B基因第3外显子存在多聚丙氨酸重复扩展变异,基因型为20/25。患儿入院后采用无创通气,睡眠时呼吸浅慢、微弱,伴血氧下降;血气分析提示二氧化碳潴留。随后改用夜间无创通气、降肺压药物治疗。复查肺动脉压力明显下降,生命体征稳定。随访至24月龄,夜间只需较低压力水平的无创通气,生长发育无异常。结论对于不明原因的肺动脉高压伴撤机困难患儿,需警惕CCHS。疑诊者应尽早针对CCHS相关基因进行靶向捕获二代测序及PHOX 2 B基因Sanger法验证。早期给予无创通气有望改善预后。     

生物瓣膜失功能原因分析

目的对植入人体内1年以上损坏的生物瓣膜作损坏原因的临床分析。方法1993年6月至2004年6月,共41例(47只)生物瓣膜失功能病人行再次手术换瓣。两次换瓣间隔时间1~16年,平均(9.7±4.1)年。对失功能生物瓣膜进行标本观察及病理学检测。结果47只带支架生物瓣膜中,瓣架无一损坏,42只瓣膜发生瓣叶撕裂或穿孔,35只有不同程度的钙化现象,且在瓣环表面和瓣脚周围有纤维组织增生,2只瓣膜有赘生物及血栓形成;47只生物瓣膜中损坏较重的瓣膜送病理学检查发现大量吞噬细胞和单核细胞浸润。结论生物瓣膜早期损坏以破损撕裂为主,晚期以钙化和破损撕裂等多种因素的复合病损结果出现。     

运动训练联合基因治疗对肾性高血压大鼠肾功能的影响

目的观察运动训练联合β1肾上腺素能受体基因治疗对肾性高血压大鼠血压、肾功能、肾脏前肾素原mRNA、肾脏β1受体mRNA和蛋白的影响，探讨其改善肾功能的机制。方法两肾一夹法制作肾性高血压模型，基因治疗采用经鼠尾静脉注射阳离子脂质体与β1反义寡核苷酸方法。检测大鼠血压、肾功能变化。半定量RT—PCR测定肾脏β1受体mRNA、前肾素原mRNA水平。Western印迹法测检肾脏β1受体的蛋白水平。结果与模型组比较，运动联合基因治疗可使血压下降并维持4周，血压下降最高达41mmHg；尿蛋白量[（45．82±6．56）比（29．12±5．22）mg／L，P〈0．01】、BUN[（13．10±2．62）比（9．05±1．84）mmol／L，P〈0．05]显著降低（P〈0．01，P〈0．05）；内生肌酐清除率显著升高（P〈0．01）；前肾素原mRNA、β1受体mRNA、蛋白表达水平显著降低（P〈0．05）。结论运动训练联合β1受体反义基因治疗可以明显地降低血压，改善肾功能；且运动训练可以增强基因治疗对β1受体mRNA和蛋白的抑制作用，在转录和翻译水平抑制过度激活的β1受体的表达。