Exploring LRP and Grad-CAM visualization to interpret multi-label-multi-class pathology prediction using chest radiography

Alam, Mahbub UlStockholms universitet,Institutionen för data- och systemvetenskap (author)

The paper won the ‘best student paper award’ provided by the IEEE Technical Committee on Computational Life Science (TCCLS). For details please visit https://2022.cbms-conference.org/awards/
The area of interpretable deep neural networks has received increased attention in recent years due to the need for transparency in various fields, including medicine, healthcare, stock market analysis, compliance with legislation, and law. Layer-wise Relevance Propagation (LRP) and Gradient-weighted Class Activation Mapping (Grad-CAM) are two widely used algorithms to interpret deep neural networks. In this work, we investigated the applicability of these two algorithms in the sensitive application area of interpreting chest radiography images. In order to get a more nuanced and balanced outcome, we use a multi-label classification-based dataset and analyze the model prediction by visualizing the outcome of LRP and Grad-CAM on the chest radiography images. The results show that LRP provides more granular heatmaps than Grad-CAM when applied to the CheXpert dataset classification model. We posit that this is due to the inherent construction difference of these algorithms (LRP is layer-wise accumulation, whereas Grad-CAM focuses primarily on the final sections in the model's architecture). Both can be useful for understanding the classification from a micro or macro level to get a superior and interpretable clinical decision support system.

Baldvinsson, Jón RúnarSkatturinn (Iceland Revenue and Customs), Reykjavík, Iceland (author)
Wang, YuxiaQamcom Research and Technology, Stockholm, Sweden (author)
Stockholms universitetInstitutionen för data- och systemvetenskap (creator_code:org_t)

In:2022 IEEE 35th International Symposium on Computer-Based Medical Systems (CBMS): IEEE conference proceedings, s. 258-2639781665467704

About the subject

NATURAL SCIENCES: NATURAL SCIENCES; and Computer and Inf ...; and Computer Science ...

Kungliga biblioteket hanterar dina personuppgifter i enlighet med EU:s dataskyddsförordning (2018), GDPR. Läs mer om hur det funkar här.
Så här hanterar KB dina uppgifter vid användning av denna tjänst.