Lesion segmentation using entropy based membership
(2018) 2nd International Conference on Electronics, Materials Engineering and Nano-Technology, IEMENTech 2018- Abstract
The challenges faced in medical image processing makes image processing domain one of the active research areas nowadays. Brain tumors are aberrant growth formed by cells reproducing in an abnormal manner. Detecting the location of brain lesion (tumor) is essential as it helps in the diagnosis and cure of tumor. This paper proposes a novel clustering approach for the brain lesions in BRATS training 2017. This work is an advancement of a well-known Fuzzy C-Means (FCM) method to separate out different tissues of MR Brain images. Here, a novel membership function has been proposed. The evaluation of the adopted approach is compared using several validity functions with other state-of-the-art methods. The computational result on brain MR... (More)
The challenges faced in medical image processing makes image processing domain one of the active research areas nowadays. Brain tumors are aberrant growth formed by cells reproducing in an abnormal manner. Detecting the location of brain lesion (tumor) is essential as it helps in the diagnosis and cure of tumor. This paper proposes a novel clustering approach for the brain lesions in BRATS training 2017. This work is an advancement of a well-known Fuzzy C-Means (FCM) method to separate out different tissues of MR Brain images. Here, a novel membership function has been proposed. The evaluation of the adopted approach is compared using several validity functions with other state-of-the-art methods. The computational result on brain MR image shows that the proposed technique is powerful and precise than the standard FCM process. This proposed approach, based on entropy based variation of fuzzy membership function is able to handle uncertainty and vagueness in images by adopting new distance modification to calculate the optimum cluster for even mixed pixels. The method has been tested on BRATS 2017 training dataset. Compared to K-Means, K-Medoids and Fuzzy C-Means the proposed method significantly reduces the segmentation errors.
(Less)
- author
- Bose, Ananya ; Sarkar, Anasua LU and Maulik, Ujjwal
- publishing date
- 2018-09-13
- type
- Chapter in Book/Report/Conference proceeding
- publication status
- published
- subject
- keywords
- BRATS 2017, Clustering, Entropy, FCM, Gaussian membership, K-means, K-medoids, Tumor
- host publication
- 2018 2nd International Conference on Electronics, Materials Engineering and Nano-Technology, IEMENTech 2018
- editor
- Sarkar, Mili ; Chakrabarty, Ratna ; Taki, G. S. and Chakrabarti, Satyajit
- article number
- 8465342
- publisher
- IEEE - Institute of Electrical and Electronics Engineers Inc.
- conference name
- 2nd International Conference on Electronics, Materials Engineering and Nano-Technology, IEMENTech 2018
- conference location
- Science City, Kolkata, India
- conference dates
- 2018-04-04 - 2018-04-05
- external identifiers
-
- scopus:85054512670
- ISBN
- 978-1-5386-5550-4
- 9781538655498
- DOI
- 10.1109/IEMENTECH.2018.8465342
- language
- English
- LU publication?
- no
- id
- 881ae061-0e75-4078-8389-9d0af226ad63
- date added to LUP
- 2019-04-02 09:43:39
- date last changed
- 2024-06-11 07:51:38
@inproceedings{881ae061-0e75-4078-8389-9d0af226ad63, abstract = {{<p>The challenges faced in medical image processing makes image processing domain one of the active research areas nowadays. Brain tumors are aberrant growth formed by cells reproducing in an abnormal manner. Detecting the location of brain lesion (tumor) is essential as it helps in the diagnosis and cure of tumor. This paper proposes a novel clustering approach for the brain lesions in BRATS training 2017. This work is an advancement of a well-known Fuzzy C-Means (FCM) method to separate out different tissues of MR Brain images. Here, a novel membership function has been proposed. The evaluation of the adopted approach is compared using several validity functions with other state-of-the-art methods. The computational result on brain MR image shows that the proposed technique is powerful and precise than the standard FCM process. This proposed approach, based on entropy based variation of fuzzy membership function is able to handle uncertainty and vagueness in images by adopting new distance modification to calculate the optimum cluster for even mixed pixels. The method has been tested on BRATS 2017 training dataset. Compared to K-Means, K-Medoids and Fuzzy C-Means the proposed method significantly reduces the segmentation errors.</p>}}, author = {{Bose, Ananya and Sarkar, Anasua and Maulik, Ujjwal}}, booktitle = {{2018 2nd International Conference on Electronics, Materials Engineering and Nano-Technology, IEMENTech 2018}}, editor = {{Sarkar, Mili and Chakrabarty, Ratna and Taki, G. S. and Chakrabarti, Satyajit}}, isbn = {{978-1-5386-5550-4}}, keywords = {{BRATS 2017; Clustering; Entropy; FCM; Gaussian membership; K-means; K-medoids; Tumor}}, language = {{eng}}, month = {{09}}, publisher = {{IEEE - Institute of Electrical and Electronics Engineers Inc.}}, title = {{Lesion segmentation using entropy based membership}}, url = {{http://dx.doi.org/10.1109/IEMENTECH.2018.8465342}}, doi = {{10.1109/IEMENTECH.2018.8465342}}, year = {{2018}}, }