AI-based improvement in lung cancer detection on chest radiographs: results of a multi-reader study in NLST dataset.

OBJECTIVE: Assess if deep learning-based artificial intelligence (AI) algorithm improves reader performance for lung cancer detection on chest X-rays (CXRs). METHODS: This reader study included 173 images from cancer-positive patients (n = 98) and 346 images from cancer-negative patients (n = 196) selected from National Lung Screening Trial (NLST). Eight readers, including three radiology residents, and five board-certified radiologists, participated in the observer performance test. AI algorithm provided image-level probability of pulmonary nodule or mass on CXRs and a heatmap of detected lesions. Reader performance was compared with AUC, sensitivity, specificity, false-positives per image (FPPI), and rates of chest CT recommendations. RESULTS: With AI, the average sensitivity of readers for the detection of visible lung cancer increased for residents, but was similar for radiologists compared to that without AI (0.61 [95% CI, 0.55-0.67] vs. 0.72 [95% CI, 0.66-0.77], p = 0.016 for residents, and 0.76 [95% CI, 0.72-0.81] vs. 0.76 [95% CI, 0.72-0.81, p = 1.00 for radiologists), while false-positive findings per image (FPPI) was similar for residents, but decreased for radiologists (0.15 [95% CI, 0.11-0.18] vs. 0.12 [95% CI, 0.09-0.16], p = 0.13 for residents, and 0.24 [95% CI, 0.20-0.29] vs. 0.17 [95% CI, 0.13-0.20], p < 0.001 for radiologists). With AI, the average rate of chest CT recommendation in patients positive for visible cancer increased for residents, but was similar for radiologists (54.7% [95% CI, 48.2-61.2%] vs. 70.2% [95% CI, 64.2-76.2%], p < 0.001 for residents and 72.5% [95% CI, 68.0-77.1%] vs. 73.9% [95% CI, 69.4-78.3%], p = 0.68 for radiologists), while that in cancer-negative patients was similar for residents, but decreased for radiologists (11.2% [95% CI, 9.6-13.1%] vs. 9.8% [95% CI, 8.0-11.6%], p = 0.32 for residents and 16.4% [95% CI, 14.7-18.2%] vs. 11.7% [95% CI, 10.2-13.3%], p < 0.001 for radiologists). CONCLUSIONS: AI algorithm can enhance the performance of readers for the detection of lung cancers on chest radiographs when used as second reader. KEY POINTS: • Reader study in the NLST dataset shows that AI algorithm had sensitivity benefit for residents and specificity benefit for radiologists for the detection of visible lung cancer. • With AI, radiology residents were able to recommend more chest CT examinations (54.7% vs 70.2%, p < 0.001) for patients with visible lung cancer. • With AI, radiologists recommended significantly less proportion of unnecessary chest CT examinations (16.4% vs. 11.7%, p < 0.001) in cancer-negative patients.

Quantitative Framework for Risk Stratification of Thyroid Nodules With Ultrasound: A Step Toward Automated Triage of Thyroid Cancer.

OBJECTIVE. The purpose of this study was to explore whether a quantitative framework can be used to sonographically differentiate benign and malignant thyroid nodules at a level comparable to that of experts. MATERIALS AND METHODS. A dataset of ultrasound images of 92 biopsy-confirmed nodules was collected retrospectively. The nodules were delineated and annotated by two expert radiologists using the standardized Thyroid Imaging Reporting and Data System lexicon of the American College of Radiology. In the framework studied, quantitative features of echogenicity, texture, edge sharpness, and margin curvature properties of thyroid nodules were analyzed in a regularized logistic regression model to predict malignancy of a nodule. The framework was validated by leave-one-out cross-validation technique, and ROC AUC, sensitivity, and specificity were used to compare with those obtained with six expert annotation-based classifiers. RESULTS. The AUC of the proposed method was 0.828 (95% CI, 0.715-0.942), which was greater than or comparable to that of the expert classifiers, for which the AUC values ranged from 0.299 to 0.829 (p = 0.99). Use of the proposed framework could have avoided biopsy of 20 of 46 benign nodules in a curative strategy (at sensitivity of 1, statistically significantly higher than three expert classifiers) or helped identify 10 of 46 malignancies in a conservative strategy (at specificity of 1, statistically significantly higher than five expert classifiers). CONCLUSION. When the proposed quantitative framework was used, thyroid nodule malignancy was predicted at the level of expert classifiers. Such a framework may ultimately prove useful as the basis for a fully automated system of thyroid nodule triage.

Central echogenic areas in thyroid nodules: Diagnostic performance in prediction of papillary cancer.

PURPOSE: To determine the diagnostic performance of the "central echogenic area" sonographic finding in differentiating papillary carcinomas from benign nodules and to how this finding may be used to improve fine needle aspiration(FNA) technique/utilization. MATERIALS AND METHODS: We retrospectively analyzed ultrasound guided FNAs of thyroid nodules between 1 and 3 cm for central echogenic areas. 92 patients (evenly distributed benign vs papillary carcinoma) were evaluated by a blinded reader for areas of non-shadowing homogenously echogenic centers within the nodules and correlated with FNA proven pathologic diagnosis. A selection of nodules with the central echogenic area finding were selected for further slide review to establish a pathologic basis for the finding. RESULTS: Diagnostic performance of the "central echogenic area" feature in papillary thyroid cancers was 52.2% sensitive and 91.3% specific for papillary thyroid carcinoma with a PPV of 85.7% and NPV of 65.6%. There was a significant correlation with a p < 0.01 between the central echogenic area finding and papillary carcinoma. On pathologic slide review, nodules with central echogenic areas consistently demonstrated a central scar with conglomerate fibrosis and very few viable cells. CONCLUSION: Despite its relatively low sensitivity, the central echogenic area finding is highly specific for papillary carcinoma of the thyroid and can be a useful sonographic finding in decisions regarding FNA. Additionally, due to the paucity of cells and high density of conglomerate fibrosis, central echogenic areas should be avoided during FNA to decrease the chance of an inadequate sample collection.

Diagnostic Performance of Margin Features in Thyroid Nodules in Prediction of Malignancy.

OBJECTIVE: The purpose of this study was to evaluate thyroid nodule margins for specific morphologic features and determine the diagnostic performance of these features in differentiating papillary carcinoma from benign thyroid nodules. MATERIALS AND METHODS: Nodules measuring 1-3 cm in largest diameter that had been evaluated with high-resolution ultrasound (12-18 MHz) and ultrasound-guided biopsy with definitive pathologic diagnosis were analyzed. Three blinded board-certified readers evaluated high-resolution images of each nodule for jagged edges, lobulated borders, and curved borders along their margins. Reader interpretations were correlated with the pathologic diagnosis to determine the diagnostic performance of each feature. A board-certified pathologist analyzed 10 randomly selected nodules with jagged edges by slide review to evaluate for structural correlation with the imaging finding. RESULTS: The diagnostic performance of jagged edges in papillary carcinoma of the thyroid was 67.4% sensitive and 78.3% specific (odds ratio, 7.44; p < 0.001) for malignancy. Jagged edges correlated with infiltrative variant expansion at slide review. Lobulated borders had sensitivity of 76.1% and specificity of 60.9% for papillary carcinoma (odds ratio, 4.95; p = 0.001) for malignancy. Curved borders were not a significant predictor of papillary carcinoma. CONCLUSION: Jagged edges and lobulated borders of thyroid nodule margins are statistically significant predictors of papillary carcinoma of the thyroid. Jagged edges correlate with infiltrative-type expansion and may be useful predictors of more aggressive papillary carcinomas.

Toward Automated Pre-Biopsy Thyroid Cancer Risk Estimation in Ultrasound.

We propose a computational framework for automated cancer risk estimation of thyroid nodules visualized in ultrasound (US) images. Our framework estimates the probability of nodule malignancy using random forests on a rich set of computational features. An expert radiologist annotated thyroid nodules in 93 biopsy-confirmed patients using semantic image descriptors derived from standardized lexicon. On our dataset, the AUC of the proposed method was 0.70, which was comparable to five baseline expert annotation-based classifiers with AUC values from 0.72 to 0.81. Moreover, the use of the framework for decision making on nodule biopsy could have spared five out of 46 benign nodule biopsies at no cost to the health of patients with malignancies. Our results confirm the feasibility of computer-aided tools for noninvasive malignancy risk estimation in patients with thyroid nodules that could help to decrease the number of unnecessary biopsies and surgeries.

Combinatorial hydrogel library enables identification of materials that mitigate the foreign body response in primates.

The foreign body response is an immune-mediated reaction that can lead to the failure of implanted medical devices and discomfort for the recipient. There is a critical need for biomaterials that overcome this key challenge in the development of medical devices. Here we use a combinatorial approach for covalent chemical modification to generate a large library of variants of one of the most widely used hydrogel biomaterials, alginate. We evaluated the materials in vivo and identified three triazole-containing analogs that substantially reduce foreign body reactions in both rodents and, for at least 6 months, in non-human primates. The distribution of the triazole modification creates a unique hydrogel surface that inhibits recognition by macrophages and fibrous deposition. In addition to the utility of the compounds reported here, our approach may enable the discovery of other materials that mitigate the foreign body response.

Implantable, multifunctional, bioresorbable optics.

Advances in personalized medicine are symbiotic with the development of novel technologies for biomedical devices. We present an approach that combines enhanced imaging of malignancies, therapeutics, and feedback about therapeutics in a single implantable, biocompatible, and resorbable device. This confluence of form and function is accomplished by capitalizing on the unique properties of silk proteins as a mechanically robust, biocompatible, optically clear biomaterial matrix that can house, stabilize, and retain the function of therapeutic components. By developing a form of high-quality microstructured optical elements, improved imaging of malignancies and of treatment monitoring can be achieved. The results demonstrate a unique family of devices for in vitro and in vivo use that provide functional biomaterials with built-in optical signal and contrast enhancement, demonstrated here with simultaneous drug delivery and feedback about drug delivery with no adverse biological effects, all while slowly degrading to regenerate native tissue.