Transfer learning for the generalization of artificial intelligence in breast cancer detection: a case-control study.

BACKGROUND: Some researchers have questioned whether artificial intelligence (AI) systems maintain their performance when used for women from populations not considered during the development of the system. PURPOSE: To evaluate the impact of transfer learning as a way of improving the generalization of AI systems in the detection of breast cancer. MATERIAL AND METHODS: This retrospective case-control Finnish study involved 191 women diagnosed with breast cancer and 191 matched healthy controls. We selected a state-of-the-art AI system for breast cancer detection trained using a large US dataset. The selected baseline system was evaluated in two experimental settings. First, we examined our private Finnish sample as an independent test set that had not been considered in the development of the system (unseen population). Second, the baseline system was retrained to attempt to improve its performance in the unseen population by means of transfer learning. To analyze performance, we used areas under the receiver operating characteristic curve (AUCs) with DeLong's test. RESULTS: Two versions of the baseline system were considered: ImageOnly and Heatmaps. The ImageOnly and Heatmaps versions yielded mean AUC values of 0.82±0.008 and 0.88±0.003 in the US dataset and 0.56 (95% CI=0.50-0.62) and 0.72 (95% CI=0.67-0.77) when evaluated in the unseen population, respectively. The retrained systems achieved AUC values of 0.61 (95% CI=0.55-0.66) and 0.69 (95% CI=0.64-0.75), respectively. There was no statistical difference between the baseline system and the retrained system. CONCLUSION: Transfer learning with a small study sample did not yield a significant improvement in the generalization of the system.

Saliency of breast lesions in breast cancer detection using artificial intelligence.

The analysis of mammograms using artificial intelligence (AI) has shown great potential for assisting breast cancer screening. We use saliency maps to study the role of breast lesions in the decision-making process of AI systems for breast cancer detection in screening mammograms. We retrospectively collected mammograms from 191 women with screen-detected breast cancer and 191 healthy controls matched by age and mammographic system. Two radiologists manually segmented the breast lesions in the mammograms from CC and MLO views. We estimated the detection performance of four deep learning-based AI systems using the area under the ROC curve (AUC) with a 95% confidence interval (CI). We used automatic thresholding on saliency maps from the AI systems to identify the areas of interest on the mammograms. Finally, we measured the overlap between these areas of interest and the segmented breast lesions using Dice's similarity coefficient (DSC). The detection performance of the AI systems ranged from low to moderate (AUCs from 0.525 to 0.694). The overlap between the areas of interest and the breast lesions was low for all the studied methods (median DSC from 4.2% to 38.0%). The AI system with the highest cancer detection performance (AUC = 0.694, CI 0.662-0.726) showed the lowest overlap (DSC = 4.2%) with breast lesions. The areas of interest found by saliency analysis of the AI systems showed poor overlap with breast lesions. These results suggest that AI systems with the highest performance do not solely rely on localized breast lesions for their decision-making in cancer detection; rather, they incorporate information from large image regions. This work contributes to the understanding of the role of breast lesions in cancer detection using AI.

Image retrieval-based parenchymal analysis for breast cancer risk assessment.

PURPOSE: This research on breast cancer risk assessment aims to develop models that predict the likelihood of breast cancer. In recent years, the computerized analysis of visual texture patterns in mammograms, namely parenchymal analysis, has shown great potential for risk assessment. However, the visual complexity and heterogeneity of visual patterns limit the performance of parenchymal analysis in large populations. In this work, we propose a method to create individualized risk assessment models based on the radiological visual appearance (radiomic phenotypes) of the mammograms. METHODS: We developed a content-based image retrieval system to stratify mammographic analysis according to the similarities of their radiomic phenotypes. We collected 1144 mammograms from 286 women following a case-control study design. We compared the classical parenchymal analysis with the proposed approach using the area under the ROC curve (AUC) with 95% confidence intervals (CI). Statistical significance was assessed using DeLong's test ( p < $p<$ 0.05). RESULTS: At a patient level, AUC values of 0.504 (95% CI: 0.398-0.611) with classical parenchymal analysis increased to 0.813 (95% CI: 0.734-0.892) when the radiomic phenotypes are incorporated with the proposed method. In risk estimation from individual, standard mammographic views, the highest performance was obtained with the mediolateral oblique view of the right breast (RMLO), with an AUC value of 0.727 (95% CI: 0.634-0.820). Differences in performance among views were statistically significant ( p < 0.05 $p<0.05$ ) CONCLUSIONS: These results indicate that the utilization of radiomic phenotypes increases the performance of computerized risk assessment based on parenchymal analysis of mammographic images. SIGNIFICANCE: The creation of individualized risk assessment models may be leveraged to target personalized screening and prevention recommendations according to the person's risk.