|
Title:
|
MHA-SEPA: A COMPREHENSIVE MODEL FOR LUNG NODULE SEGMENTATION |
|
Author(s):
|
Khai Dinh Lai , Thai Hoang Le and Thanh Thuy Nguyen |
|
ISBN:
|
978-989-8704-62 |
|
Editors:
|
Paula Miranda and Pedro Isaías |
|
Year:
|
2024 |
|
Edition:
|
Single |
|
Keywords:
|
Lung Nodule Segmentation, Attention Mechanism, Multi-Head Attention |
|
Type:
|
Full |
|
First Page:
|
193 |
|
Last Page:
|
202 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
This research introduces a sophisticated technique for segmenting lung nodules in CT scans, employing the MHA-SEPA
model, which is built upon the ResUNet++ framework. The MHA-SEPA architecture combines Multi-Head Attention
mechanisms with SEBlock and Position Attention approaches to improve feature extraction by giving priority to important
spatial and channel information. This method greatly enhances the precision of lung nodule segmentation, especially for
tiny and difficult nodules. We utilize an innovative technique for data augmentation by injecting Gaussian noise, which
enhances the variety and generalizability of the training dataset. Our MHA-SEPA model obtains a remarkable Dice
Similarity Coefficient (DSC) of 89.96% based on the experimental findings obtained from the LUNA16 dataset. This work
offers a strong foundation for improving the precision and responsiveness of automated lung nodule detection, which might
possibly assist in the timely identification and management of lung cancer. |
|
|
|
|
|
|
