TY - GEN
T1 - Advancing Early-Stage Brain Tumor Detection with Segmentation by Modified_UNet
AU - Musthafa, Namya
AU - Masud, Mohammad Mehedy
AU - Memon, Qurban
N1 - Publisher Copyright:
© 2024 Copyright held by the owner/author(s)
PY - 2024/5/17
Y1 - 2024/5/17
N2 - Prompt and accurate detection of brain tumors is crucial to implement effective medical intervention, as brain tumors present a substantial health hazard. The goal of this work is to employ Magnetic Resonance Imaging (MRI) segmentation for brain tumor identification. A robust and efficient strategy is proposed to address the pressing need for more precise procedures in the diagnosis of brain tumors. The proposed technique makes use of the pre-trained UNet model backbone, which is intended to improve segmentation performance. The UNet encoder incorporates pre-trained models as backbone, thereby augmenting feature representation diversity and resulting in improved segmentation accuracy. Empirical evaluations of the dataset show promising results in brain tumor detection compared to existing approaches. An intersection over union (IoU) score of 0.8258 is achieved by this approach over a dataset of 6128 MRI slices with the threshold set to 0.9 which is relatively higher than UNet segmentation.
AB - Prompt and accurate detection of brain tumors is crucial to implement effective medical intervention, as brain tumors present a substantial health hazard. The goal of this work is to employ Magnetic Resonance Imaging (MRI) segmentation for brain tumor identification. A robust and efficient strategy is proposed to address the pressing need for more precise procedures in the diagnosis of brain tumors. The proposed technique makes use of the pre-trained UNet model backbone, which is intended to improve segmentation performance. The UNet encoder incorporates pre-trained models as backbone, thereby augmenting feature representation diversity and resulting in improved segmentation accuracy. Empirical evaluations of the dataset show promising results in brain tumor detection compared to existing approaches. An intersection over union (IoU) score of 0.8258 is achieved by this approach over a dataset of 6128 MRI slices with the threshold set to 0.9 which is relatively higher than UNet segmentation.
KW - Backbone model
KW - early-stage brain tumor detection
KW - Encoder-Decoder
KW - feature extraction
KW - ResNet
KW - Segmentation
KW - UNet
UR - http://www.scopus.com/inward/record.url?scp=85204619002&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85204619002&partnerID=8YFLogxK
U2 - 10.1145/3673971.3674001
DO - 10.1145/3673971.3674001
M3 - Conference contribution
AN - SCOPUS:85204619002
T3 - ACM International Conference Proceeding Series
SP - 52
EP - 57
BT - ICMHI 2024 - 2024 8th International Conference on Medical and Health Informatics
PB - Association for Computing Machinery
T2 - 8th International Conference on Medical and Health Informatics, ICMHI 2024
Y2 - 17 May 2024 through 19 May 2024
ER -