{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/a-novel-focal-tversky-loss-function-with","title":"A Novel Focal Tversky loss function with improved Attention U-Net for lesion segmentation","arxiv_id":"1810.07842","date":"2018-10-18","proceeding":null,"authors":["Nabila Abraham","Naimul Mefraz Khan"],"abstract":"We propose a generalized focal loss function based on the Tversky index to\naddress the issue of data imbalance in medical image segmentation. Compared to\nthe commonly used Dice loss, our loss function achieves a better trade off\nbetween precision and recall when training on small structures such as lesions.\nTo evaluate our loss function, we improve the attention U-Net model by\nincorporating an image pyramid to preserve contextual features. We experiment\non the BUS 2017 dataset and ISIC 2018 dataset where lesions occupy 4.84% and\n21.4% of the images area and improve segmentation accuracy when compared to the\nstandard U-Net by 25.7% and 3.6%, respectively.","url_abs":"http://arxiv.org/abs/1810.07842v1","url_pdf":"http://arxiv.org/pdf/1810.07842v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"a-novel-focal-tversky-loss-function-with","repo_url":"https://github.com/nabsabraham/focal-tversky-unet","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"a-novel-focal-tversky-loss-function-with","repo_url":"https://github.com/EvgenyDyshlyuk/Oil_Seep_Detection","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"a-novel-focal-tversky-loss-function-with","repo_url":"https://github.com/Jo-dsa/SemanticSeg","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"a-novel-focal-tversky-loss-function-with","repo_url":"https://github.com/SahinTiryaki/Brain-tumor-segmentation-Vgg19UNet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"a-novel-focal-tversky-loss-function-with","repo_url":"https://github.com/woans0104/project_review","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}},{"paper_slug":"a-novel-focal-tversky-loss-function-with","repo_url":"https://github.com/woans0104/sk_project","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"image-segmentation","task_name":"Image Segmentation"},{"task_slug":"lesion-segmentation","task_name":"Lesion Segmentation"},{"task_slug":"medical-image-segmentation","task_name":"Medical Image Segmentation"},{"task_slug":"segmentation","task_name":"Segmentation"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"},{"task_slug":"skin-lesion-segmentation","task_name":"Skin Lesion Segmentation"}],"methods":[{"method_slug":"concatenated-skip-connection","method_name":"Concatenated Skip Connection"},{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"focal-loss","method_name":"Focal Loss"},{"method_slug":"max-pooling","method_name":"Max Pooling"},{"method_slug":"relu","method_name":"ReLU"},{"method_slug":"u-net","method_name":"U-Net"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/lesion-segmentation-on-bus-2017-dataset-b","task":"Lesion Segmentation","dataset":"BUS 2017 Dataset B","model":"Attn U-Net + Multi-Input + FTL","rank_in_archive_order":1,"of":4,"metrics":{"Dice Score":"0.804"},"uses_additional_data":false},{"leaderboard":"/sota/lesion-segmentation-on-bus-2017-dataset-b","task":"Lesion Segmentation","dataset":"BUS 2017 Dataset B","model":"U-Net + FTL","rank_in_archive_order":3,"of":4,"metrics":{"Dice Score":"0.669"},"uses_additional_data":false},{"leaderboard":"/sota/lesion-segmentation-on-bus-2017-dataset-b","task":"Lesion Segmentation","dataset":"BUS 2017 Dataset B","model":"Attn U-Net + DL","rank_in_archive_order":4,"of":4,"metrics":{"Dice Score":"0.615"},"uses_additional_data":false},{"leaderboard":"/sota/lesion-segmentation-on-isic-2018","task":"Lesion Segmentation","dataset":"ISIC 2018","model":"Attn U-Net + Multi-Input + FTL","rank_in_archive_order":12,"of":17,"metrics":{"mean Dice":"0.856"},"uses_additional_data":false},{"leaderboard":"/sota/lesion-segmentation-on-isic-2018","task":"Lesion Segmentation","dataset":"ISIC 2018","model":"U-Net + FTL","rank_in_archive_order":15,"of":17,"metrics":{"mean Dice":"0.829"},"uses_additional_data":false},{"leaderboard":"/sota/lesion-segmentation-on-isic-2018","task":"Lesion Segmentation","dataset":"ISIC 2018","model":"Attn U-Net + DL","rank_in_archive_order":16,"of":17,"metrics":{"mean Dice":"0.806"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/1810.07842","atlas_url":"https://app.syntology.ai/?focus=1810.07842","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}