| 1. |
Waldamichael F.G., Yermakova A., Żołek N., On Influence of Size Markers in Ultrasound Images on Breast Lesion Segmentation Accuracy,
AIME, International Conference on Artificial Intelligence in Medicine, 2025-07-09/07-12, Salt Lake City (Utah) (US), pp.1-5, 2025 Streszczenie:
In the pursuit of advancing machine learning application
in medical diagnostics, the integrity of data plays a pivotal role. This
study delves into the influence of size markers in sonomammography on
the segmentation quality of breast lesions. Utilizing a dataset predominantly
composed of images with measurement markers, we meticulously
removed these markers to evaluate their influence on the performance of
neural networks. Our investigation employed ELUnet and DeepLabV3
segmentation networks, trained and tested across various configurations
to discern the effects of marker presence and giving the discrepancies at
a level reaching 8% Słowa kluczowe:
Sonomammography , Machine Learning , Lesion Segmentation , Data quality Afiliacje autorów:
| Waldamichael F.G. | - | IPPT PAN | | Yermakova A. | - | IPPT PAN | | Żołek N. | - | IPPT PAN |
| | 70p. |
