Hierarchical combinatorial deep learning architecture for pancreas segmentation of medical computed tomography cancer images

Abstract

BackgroundEfficient computational recognition and segmentation of target organ from medical images are foundational in diagnosis and treatment, especially about pancreas cancer. In practice, the diversity in appearance of pancreas and organs in abdomen, makes detailed texture information of objects important in segmentation algorithm. According to our observations, however, the structures of previous networks, such as the Richer Feature Convolutional Network (RCF), are too coarse to segment the object (pancreas) accurately, especially the edge.MethodIn this paper, we extend the RCF, proposed to the field of edge detection, for the challenging pancreas segmentation, and put forward a novel pancreas segmentation network. By employing multi-layer up-sampling structure replacing the simple up-sampling operation in all stages, the proposed network fully considers the multi-scale detailed contexture information of object (pancreas) to perform per-pixel segmentation. Additionally, using the CT scans, we supply and train our network, thus get an effective pipeline.ResultWorking with our pipeline with multi-layer up-sampling model, we achieve better performance than RCF in the task of single object (pancreas) segmentation. Besides, combining with multi scale input, we achieve the 76.36% DSC (Dice Similarity Coefficient) value in testing data.ConclusionThe results of our experiments show that our advanced model works better than previous networks in our dataset. On the other words, it has better ability in catching detailed contexture information. Therefore, our new single object segmentation model has practical meaning in computational automatic diagnosis.

Evaluation of deep learning‐based auto‐segmentation algorithms for delineating clinical target volume and organs at risk involving data for 125 cervical cancer patients

“…In terms of time requirements, the average time for delineation of a case, including the data pre-processing and post-processing great application potential in disease diagnosis, [44][45][46][47] lesion recognition, [48][49][50][51][52] and image segmentation. 28,[52][53][54] Especially in image segmentation, accuracy has always been the focus of attention. Most previous studies developed several superior models to improve the accuracy of auto-segmentation, and some studies compared the accuracy of different auto-segmentation models.…”

“…In recent years, an increasing number of deep learning‐based methods have been applied to the field of medical imaging, which have great application potential in disease diagnosis, 44‐47 lesion recognition, 48‐52 and image segmentation 28,52–54 . Especially in image segmentation, accuracy has always been the focus of attention.…”

Image Dehazing Based on (CMTnet) Cascaded Multi-scale Convolutional Neural Networks and Efficient Light Estimation Algorithm

“…This kind of scientific revolution grabs the attention of many researchers to adopt it for solving several low-level computer vision tasks, e.g., image segmentation [21][22][23], object detection [24], image denoising [25], image restoration [33], and image enhancement [34]. Recently, image dehazing has become a controversial topic that opens a challenge to many researchers in image processing and computer vision fields.…”

“…At present, Convolutional Neural Networks (CNNs) have attained great success in addressing several low-level computer vision and image processing tasks, e.g., image segmentation [21][22][23], object detection [24], and image denoising [25]. Many researchers [26][27][28][29] have applied Convolutional Neural Networks (CNNs) to explore haze-relevant features deeply.…”

TumorGAN: A Multi-Modal Data Augmentation Framework for Brain Tumor Segmentation

“…Generally, multi-modal data can lead to a better performance result, as compared to approaches based on a single modality, because more information about the tumor could be captured by different imaging methods [6]. Motivated by the success of deep learning, researchers soon applied deep neural networks to solve various medical imaging-related problems [7][8][9]. However, unlike classification, labeling medical images for segmentation is challenging, as it is time-consuming and requires medical specialists [10].…”