Titolo del convegno/congresso: ICPR 2022 International Conference on Pattern Recognition - Artificial Intelligence for Healthcare Applications 2nd International Workshop, 21-25/08/2022.
Autori: Gianluca Carloni (1,2,*), Andrea Berti (1,2,*), Chiara Iacconi (3), Maria Antonietta Pascali (1), Sara Colantonio (1)
Affiliazioni: (1): Institute of Information Science and Technologies (ISTI), National Research Council of Italy (CNR), Pisa, Italy. (2): Department of Information Engineering, University of Pisa, Pisa, Italy. (3): UOSD breast radiology, territorial area of Massa Carrara, Azienda USL Toscana Nord-Ovest, Carrara, Italy. *: These authors share the first authorship.
Abstract:
| Deep learning models have become state-of-the-art in many areas, ranging from computer vision to agriculture research. However, concerns have been raised with respect to the transparency of their decisions, especially in the image domain. In this regard, Explainable Artificial Intelligence has been gaining popularity in recent years. The ProtoPNet model, which breaks down an image into prototypes and uses evidence gathered from the prototypes to classify an image, represents an appealing approach. Still, questions regarding its effectiveness arise when the application domain changes from real-world natural images to gray-scale medical images. This work explores the applicability of prototypical part learning in medical imaging by experimenting with ProtoPNet on a breast masses classification task. The two considered aspects were the classification capabilities and the validity of explanations. We looked for the optimal model's hyperparameter configuration via a random search. We trained the model in a five-fold CV supervised framework, with mammogram images cropped around the lesions and ground-truth labels of benign/malignant masses. Then, we compared the performance metrics of ProtoPNet to that of the corresponding base architecture, which was ResNet18, trained under the same framework. In addition, an experienced radiologist provided a clinical viewpoint on the quality of the learned prototypes, the patch activations, and the global explanations. We achieved a Recall of 0.769 and an area under the receiver operating characteristic curve of 0.719 in our experiments. Even though our findings are non-optimal for entering the clinical practice yet, the radiologist found ProtoPNet's explanations very intuitive, reporting a high level of satisfaction. Therefore, we believe that prototypical part learning offers a reasonable and promising trade-off between classification performance and the quality of the related explanation. |
Pagine: 19
Rivista: Lecture notes in computer science (Internet), Attiva dal 1973, Editore: Springer International - Cham, Paese di pubblicazione: Svizzera Lingua: inglese ISSN: 1611-3349.
DOI: https://doi.org/10.1007/978-3-031-37660-3_38
https://link.springer.com/chapter/10.1007/978-3-031-37660-3_38
