To Be (B Read) or Not to Be (B Read), That Is the Question: Conflict of Interest and Radiographic Interpretation

Redlich, Carrie A.; Rosenman, Kenneth D.

doi:10.1513/annalsats.202107-764ed

Cited by 1 publication

(2 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For the deep learning architecture, motivated by the ordinal nature of Pneumoconioses grading, we model this problem as a multi-task learning scenario. Instead of treating it as a conventional four-class problem, we decompose it into three binary classification tasks: distinguishing between class 0 and the higher classes (1,2,3), between class 1 and the higher classes (2,3), and between class 2 and class 3. This approach allows for the incorporation of the ordinal relationship between the classes into the model architecture.…”

Section: Combined Features With Resnet On Multi-task Lossmentioning

confidence: 99%

“…Both inter-and intra-reader variability were major concerns that led the International Labor Organization (ILO) and the NIOSH to standardize the classification of radiographs for the pneumoconioses and to develop the NIOSH B Reader Certification Program in 1974. While the program has improved the proficiency of the certified physicians, there remain key challenges including: a) the limited numbers of certified B readers, currently only 209 in the U.S., 2 b) continuing concerns about inter-and intra-reader variability, 3 and c) concerns about the influence of financial conflicts of interests. For these reasons, there is a pressing need for artificial intelligence (AI) algorithms to ensure the objective and consistent radiographic assessment of the pneumoconioses.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Deep learning methods for multi-class pneumoconioses grading of chest radiographs

Liu,

Loveless,

Huang

et al. 2024

Medical Imaging 2024: Computer-Aided Diagnosis

View full text Add to dashboard Cite

This study proposes different deep learning approaches to automatically classify pneumoconiosis based on the International Labour Office ("ILO") classification system. Through collaboration with the National Institute for Occupational Safety and Health (NIOSH), this study curated a custom dataset of chest radiographs with (N=520) and without (N=149) pneumoconiosis. The four-point major category scale of profusion (concentration) of small opacities (0, 1, 2, or 3) were considered in this study. Several deep learning models were evaluated for classifying different levels of profusion grades including: 1) Transfer learning with models pre-trained with public chest radiograph repositories, 2) Hand-crafted radiomic feature extraction, and 3) Hybrid architecture that integrates hand-crafted radiomic features with transfer-learned features using a loss function that incorporates the inherent ordinality within the profusion grade scale. For profusion grade 0 vs. grade 3, both the transfer learning and radiomic feature extraction methods obtained test-set accuracies of greater than 91%. The highest prediction accuracy for normal (profusion grade 0) vs. abnormal (profusion grade 1, 2 and 3) was 83% using the transfer learning method. Under the setting of multi-class identification of four profusion grades, ResNet model adapted to a ordinal multi-task loss function notably outperforms traditional models reliant on cross-entropy loss (with accuracy 58%) and achieves an accuracy of approximately 61%. The amalgamation of radiomic and ResNetderived features, coupled with the application of multi-task loss, culminates in the highest recorded accuracy of approximately 62%. This demonstrates an example case where the integration of hand-crafted and deeplearned features, along with modeling the ordinality of the classes, improves classification performance of chest radiographs.

show abstract

Section: Combined Features With Resnet On Multi-task Lossmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%