Semantic Segmentation Using Deep Learning to Extract Total Extraocular Muscles and Optic Nerve from Orbital Computed Tomography Images

Abstract Pancreas segmentation is necessary for observing lesions, analyzing anatomical structures, and predicting patient prognosis. Therefore, various studies have designed segmentation models based on convolutional neural networks for pancreas segmentation. However, the deep learning approach is limited by a lack of data, and studies conducted on a large computed tomography dataset are scarce. Therefore, this study aims to perform deep-learning-based semantic segmentation on 1,006 participants and evaluate the automatic segmentation performance of the pancreas via four individual three-dimensional segmentation networks. In this study, we performed internal validation with 1,006 patients and external validation using the Cancer Imaging Archive (TCIA) pancreas dataset. We obtained mean precision, recall, and dice similarity coefficients of 0.869, 0.842, and 0.842, respectively, for internal validation via a relevant approach among the four deep learning networks. Using the external dataset, the deep learning network achieved mean precision, recall, and dice similarity coefficients of 0.779, 0.749, and 0.735, respectively. We expect that generalized deep-learning-based systems can assist clinical decisions by providing accurate pancreatic segmentation and quantitative information of the pancreas for abdominal computed tomography.

Download Full-text

Computed Tomography Anatomy of the Optic Canal

Otolaryngology ◽

10.1016/j.otohns.2008.05.239 ◽

2008 ◽

Vol 139 (2_suppl) ◽

pp. P74-P74 ◽

Cited By ~ 4

Author(s):

Catherine K Hart ◽

Lee A Zimmer

Keyword(s):

Computed Tomography ◽

Optic Nerve ◽

Sphenoid Sinus ◽

Optic Canal ◽

Canal Wall ◽

Nerve Decompression ◽

Computed Tomography Images ◽

Optic Nerve Decompression ◽

Wide Range ◽

Potential Area

Objective (1) Analyze the radiographic anatomy of the optic canal in relationship to the sphenoid sinus. (2) Understand the role variation in optic canal anatomy may have in the variability of outcomes in optic nerve decompression. Methods Fine cut computed tomography images of the sinuses were obtained with an IRB waiver. Optic canal dimensions were measured on sinus computed tomography images of 96 patients. 191 optic canals were analyzed (111 females, 80 males). Student T-test calculations were performed for statistical analysis on computer software. Results The average medial canal wall length was 1.48 centimeters (range 0.7–2.3). The length in males was 1.61 centimeters (1.1–2.3) as compared to 1.39 centimeters (0.7–2.0) in females (p=8.0–7). The average degree of exposure of the optic canal exposed to the sphenoid sinus was 101.3 degrees (56–176). The degree of exposure was 105.6 in males versus 98.2 in females (p=.01). The potential area of canal exposed to the sphenoid sinus was 0.66 centimeters squared or 28% of the total surface area. The potential area exposed to the sphenoid sinus in males was 0.76cm2 (28%) and 0.58 centimeters squared (27%) in females. Conclusions A wide range in medial canal wall length and exposure of the bony optic canal to the sphenoid sinus exists on CT images. The variation in medial canal wall length and in optic canal exposure to the sphenoid sinus may contribute to the variability in success rates of endoscopic optic nerve decompression for optic neuropathy.

Download Full-text

Weakly supervised semantic segmentation of tomographic images in the diagnosis of stroke

Journal of Physics Conference Series ◽

10.1088/1742-6596/2099/1/012021 ◽

2021 ◽

Vol 2099 (1) ◽

pp. 012021

Author(s):

A V Dobshik ◽

A A Tulupov ◽

V B Berikov

Keyword(s):

Computed Tomography ◽

Network Architecture ◽

Semantic Segmentation ◽

Training Data ◽

Neural Network Architecture ◽

Brain Images ◽

Tomographic Images ◽

Computed Tomography Images ◽

Segmentation Accuracy ◽

Weakly Supervised

Abstract This paper presents an automatic algorithm for the segmentation of areas affected by an acute stroke in the non-contrast computed tomography brain images. The proposed algorithm is designed for learning in a weakly supervised scenario when some images are labeled accurately, and some images are labeled inaccurately. Wrong labels appear as a result of inaccuracy made by a radiologist in the process of manual annotation of computed tomography images. We propose methods for solving the segmentation problem in the case of inaccurately labeled training data. We use the U-Net neural network architecture with several modifications. Experiments on real computed tomography scans show that the proposed methods increase the segmentation accuracy.

Download Full-text

Diagnosis of COVID-19 Infection Using Three-Dimensional Semantic Segmentation and Classification of Computed Tomography Images

Computers Materials & Continua ◽

10.32604/cmc.2021.014199 ◽

2021 ◽

Vol 68 (2) ◽

pp. 2451-2467

Author(s):

Javaria Amin ◽

Muhammad Sharif ◽

Muhammad Almas Anjum ◽

Yunyoung Nam ◽

Seifedine Kadry ◽

...

Keyword(s):

Computed Tomography ◽

Three Dimensional ◽

Semantic Segmentation ◽

Computed Tomography Images

Download Full-text

A Deep Learning System for Classifying T Stage and Predicting Prognosis of Colorectal Cancer via Preoperative Computed Tomography Images

SSRN Electronic Journal ◽

10.2139/ssrn.3900665 ◽

2021 ◽

Author(s):

Xudong Zhu ◽

Yong Ren ◽

Sidong Xie ◽

Zikai Cai ◽

Zongheng Zheng ◽

...

Keyword(s):

Colorectal Cancer ◽

Computed Tomography ◽

Deep Learning ◽

Learning System ◽

T Stage ◽

Computed Tomography Images

Download Full-text

Computed Tomography Brain Images Semantic Segmentation

Advances in Multimedia and Interactive Technologies - Handbook of Research on Advanced Concepts in Real-Time Image and Video Processing ◽

10.4018/978-1-5225-2848-7.ch004 ◽

2017 ◽

pp. 77-104

Author(s):

Poonam Fauzdar ◽

Sarvesh Kumar

Keyword(s):

Computed Tomography ◽

Positive Predictive Value ◽

Euclidean Distance ◽

Global Analysis ◽

Semantic Segmentation ◽

Brain Images ◽

Minimum Euclidean Distance ◽

Computed Tomography Images ◽

Multi Level ◽

Reduced Time

In this paper we applianced an approach for segmenting brain tumour regions in a computed tomography images by proposing a multi-level fuzzy technique with quantization and minimum computed Euclidean distance applied to morphologically divided skull part. Since the edges identified with closed contours and further improved by adding minimum Euclidean distance, that is why the numerous results that are analyzed are very assuring and algorithm poses following advantages like less cost, global analysis of image, reduced time, more specificity and positive predictive value.

Download Full-text

U-Net Plus: Deep Semantic Segmentation for Esophagus and Esophageal Cancer in Computed Tomography Images

IEEE Access ◽

10.1109/access.2019.2923760 ◽

2019 ◽

Vol 7 ◽

pp. 82867-82877 ◽

Cited By ~ 6

Author(s):

Shuchao Chen ◽

Han Yang ◽

Jiawen Fu ◽

Weijian Mei ◽

Shuai Ren ◽

...

Keyword(s):

Computed Tomography ◽

Esophageal Cancer ◽

Semantic Segmentation ◽

Computed Tomography Images

Download Full-text

A 16-Year Study on Incidence and Progression of Diseased Sphenoethmoidal (Onodi) Cell

BioMed Research International ◽

10.1155/2020/2410415 ◽

2020 ◽

Vol 2020 ◽

pp. 1-5

Author(s):

Ethan I. Huang ◽

Chia-Ling Kuo ◽

Li-Wen Lee

Keyword(s):

Computed Tomography ◽

Decision Making ◽

Optic Nerve ◽

Endoscopic Sinus Surgery ◽

Visual Loss ◽

Sinus Surgery ◽

Acute Visual Loss ◽

Computed Tomography Images ◽

Onodi Cell

Traumatic operative injury of the optic nerve in an endoscopic sinus surgery may cause immediate or delayed blindness. It should be cautioned when operating in a sphenoethmoidal cell, or known as Onodi cell, with contact or bulge of the optic canal. It remains unclear how frequent progression to visual loss occurs and how long it progresses to visual loss because of a diseased sphenoethmoidal cell. Research to discuss these questions is expected to help decision making to treat diseased sphenoethmoidal cells. From July 2001 to June 2017, 216 patients received conservative endoscopic sinus surgery without opening a diseased sphenoethmoidal cell. We used their computed tomography images of paranasal sinuses to identify diseased sphenoethmoidal cells that could be associated with progression to visual loss. Among the 216 patients, 52.3% had at least one sphenoethmoidal cell, and 14.8% developed at least one diseased sphenoethmoidal cell. One patient developed acute visual loss 4412 days after the first computed tomography. Our results show that over half of the patients have a sphenoethmoidal cell but suggest a rare incidence of a diseased sphenoethmoidal cell progressing to visual loss during the follow-up period.

Download Full-text

Deep Learning Based Liver Cancer Segmentation from Computed Tomography Images

10.21203/rs.3.rs-65573/v1 ◽

2020 ◽

Author(s):

Yodit Abebe Ayalew ◽

Kinde Anlay Fante ◽

Mohammed Aliy

Keyword(s):

Computed Tomography ◽

Deep Learning ◽

Liver Cancer ◽

Ct Scan ◽

Tumor Segmentation ◽

Cancer Death ◽

Liver Segmentation ◽

Abdominal Ct ◽

Computed Tomography Images ◽

Abdominal Ct Scan

Abstract Background: Liver cancer is the sixth most common cancer worldwide. According to WHO data in 2017, the liver cancer death in Ethiopia reached 1040 (0.16%) from all cancer deaths. Hepatocellular carcinoma (HCC), primary liver cancer causes the death of around 700,000 people each year worldwide and this makes it the third leading cause of cancer death. HCC is occurred due to cirrhosis and hepatitis B or C viruses. Liver cancer mostly diagnosed with a computed tomography (CT) scan. But, the detection of the tumor from the CT scan image is difficult since tumors have similar intensity with nearby tissues and may have a different appearance depending on their type, state, and equipment setting. Nowadays deep learning methods have been used for the segmentation of liver and its tumor from the CT scan images and they are more efficient than those traditional methods. But, they are computationally expensive and need many labeled samples for training, which are difficult in the case of biomedical images. Results: A deep learning-based segmentation algorithm is employed for liver and tumor segmentation from abdominal CT scan images. Three separate UNet models, one for liver segmentation and the others two for tumor segmentation from the segmented liver and directly from the abdominal CT scan image were used. A dice score of 0.96 was obtained for liver segmentation. And a dice score of 0.74 and 0.63 was obtained for segmentation of tumor from the liver and from abdominal CT scan image respectively. Conclusion: The research improves the liver tumor segmentation that will help the physicians in the diagnosis and detection of liver tumors and in designing a treatment plan for the patient. And for the patient, it increases the patients’ chance of getting treatment and decrease the mortality rate due to liver cancer.

Download Full-text