Research

24 pages, 5735 KiB

Open AccessArticle

CI-UNet: Application of Segmentation of Medical Images of the Human Torso

by Junkang Qin, Xiao Wang, Dechang Mi, Qinmu Wu, Zhiqin He and Yu Tang

Appl. Sci. 2023, 13(12), 7293; https://doi.org/10.3390/app13127293 - 19 Jun 2023

Cited by 1 | Viewed by 1070

The study of human torso medical image segmentation is significant for computer-aided diagnosis of human examination, disease tracking, and disease prevention and treatment. In this paper, two application tasks are designed for torso medical images: the abdominal multi-organ segmentation task and the spine [...] Read more.

The study of human torso medical image segmentation is significant for computer-aided diagnosis of human examination, disease tracking, and disease prevention and treatment. In this paper, two application tasks are designed for torso medical images: the abdominal multi-organ segmentation task and the spine segmentation task. For this reason, this paper proposes a net-work model CI-UNet improve the accuracy of edge segmentation. CI-UNet is a U-shaped network structure consisting of encoding and decoding networks. Firstly, it replaces UNet’s double convolutional backbone network with a VGG16 network loaded with Transfer Learning. It feeds image information from two adjacent layers in the VGG16 network into the decoding grid via information aggregation blocks. Secondly, Polarized Self-Attention is added at the decoding network and the hopping connection, which allows the network to focus on the compelling features of the image. Finally, the image information is decoded by convolution and Up-sampling several times to obtain the segmentation results. CI-UNet was tested in the abdominal multi-organ segmentation task using the Chaos (Combined CT-MR Healthy Abdominal Organ Segmentation) open challenge dataset and compared with UNet, Attention UNet, PSPNet, DeepLabv3+ prediction networks, and dedicated network for MRI images. The experimental results showed that the average intersegmental union (mIoU) and average pixel accuracy (mPA) of organ segmentation were 82.33% and 90.10%, respectively, higher than the above comparison network. Meanwhile, we used CI-UNet for the spine dataset of the Guizhou branch of Beijing Jishuitan Hospital. The average intersegmental union (mIoU) and average pixel accuracy (mPA) of organ segmentation were 87.97% and 93.48%, respectively, which were approved by the physicians for both tasks. Full article

(This article belongs to the Special Issue Machine Learning-Based Medical Image Analysis)

► Show Figures

Figure 1

17 pages, 3862 KiB

Open AccessArticle

End-to-End Convolutional Neural Network Framework for Breast Ultrasound Analysis Using Multiple Parametric Images Generated from Radiofrequency Signals

by Soohyun Kim, Juyoung Park, Joonhwan Yi and Hyungsuk Kim

Appl. Sci. 2022, 12(10), 4942; https://doi.org/10.3390/app12104942 - 13 May 2022

Cited by 4 | Viewed by 1653

Abstract

Breast ultrasound (BUS) is an effective clinical modality for diagnosing breast abnormalities in women. Deep-learning techniques based on convolutional neural networks (CNN) have been widely used to analyze BUS images. However, the low quality of B-mode images owing to speckle noise and a [...] Read more.

Breast ultrasound (BUS) is an effective clinical modality for diagnosing breast abnormalities in women. Deep-learning techniques based on convolutional neural networks (CNN) have been widely used to analyze BUS images. However, the low quality of B-mode images owing to speckle noise and a lack of training datasets makes BUS analysis challenging in clinical applications. In this study, we proposed an end-to-end CNN framework for BUS analysis using multiple parametric images generated from radiofrequency (RF) signals. The entropy and phase images, which represent the microstructural and anatomical information, respectively, and the traditional B-mode images were used as parametric images in the time domain. In addition, the attenuation image, estimated from the frequency domain using RF signals, was used for the spectral features. Because one set of RF signals from one patient produced multiple images as CNN inputs, the proposed framework overcame the limitation of datasets in a broad sense of data augmentation while providing complementary information to compensate for the low quality of the B-mode images. The experimental results showed that the proposed architecture improved the classification accuracy and recall by 5.5% and 11.6%, respectively, compared with the traditional approach using only B-mode images. The proposed framework can be extended to various other parametric images in both the time and frequency domains using deep neural networks to improve its performance. Full article

(This article belongs to the Special Issue Machine Learning-Based Medical Image Analysis)

► Show Figures

Figure 1

23 pages, 4616 KiB

Open AccessArticle

SA-GAN: Stain Acclimation Generative Adversarial Network for Histopathology Image Analysis

by Tasleem Kausar, Adeeba Kausar, Muhammad Adnan Ashraf, Muhammad Farhan Siddique, Mingjiang Wang, Muhammad Sajid, Muhammad Zeeshan Siddique, Anwar Ul Haq and Imran Riaz

Appl. Sci. 2022, 12(1), 288; https://doi.org/10.3390/app12010288 - 29 Dec 2021

Cited by 19 | Viewed by 2673

Abstract

Histopathological image analysis is an examination of tissue under a light microscope for cancerous disease diagnosis. Computer-assisted diagnosis (CAD) systems work well by diagnosing cancer from histopathology images. However, stain variability in histopathology images is inevitable due to the use of different staining [...] Read more.

Histopathological image analysis is an examination of tissue under a light microscope for cancerous disease diagnosis. Computer-assisted diagnosis (CAD) systems work well by diagnosing cancer from histopathology images. However, stain variability in histopathology images is inevitable due to the use of different staining processes, operator ability, and scanner specifications. These stain variations present in histopathology images affect the accuracy of the CAD systems. Various stain normalization techniques have been developed to cope with inter-variability issues, allowing standardizing the appearance of images. However, in stain normalization, these methods rely on the single reference image rather than incorporate color distributions of the entire dataset. In this paper, we design a novel machine learning-based model that takes advantage of whole dataset distributions as well as color statistics of a single target image instead of relying only on a single target image. The proposed deep model, called stain acclimation generative adversarial network (SA-GAN), consists of one generator and two discriminators. The generator maps the input images from the source domain to the target domain. Among discriminators, the first discriminator forces the generated images to maintain the color patterns as of target domain. While second discriminator forces the generated images to preserve the structure contents as of source domain. The proposed model is trained using a color attribute metric, extracted from a selected template image. Therefore, the designed model not only learns dataset-specific staining properties but also image-specific textural contents. Evaluated results on four different histopathology datasets show the efficacy of SA-GAN to acclimate stain contents and enhance the quality of normalization by obtaining the highest values of performance metrics. Additionally, the proposed method is also evaluated for multiclass cancer type classification task, showing a 6.9% improvement in accuracy on ICIAR 2018 hidden test data. Full article

(This article belongs to the Special Issue Machine Learning-Based Medical Image Analysis)

► Show Figures

Figure 1

16 pages, 5405 KiB

Open AccessArticle

POCS-Augmented CycleGAN for MR Image Reconstruction

by Yiran Li, Hanlu Yang, Danfeng Xie, David Dreizin, Fuqing Zhou and Ze Wang

Appl. Sci. 2022, 12(1), 114; https://doi.org/10.3390/app12010114 - 23 Dec 2021

Viewed by 2406

Abstract

Recent years have seen increased research interest in replacing the computationally intensive Magnetic resonance (MR) image reconstruction process with deep neural networks. We claim in this paper that the traditional image reconstruction methods and deep learning (DL) are mutually complementary and can be [...] Read more.

Recent years have seen increased research interest in replacing the computationally intensive Magnetic resonance (MR) image reconstruction process with deep neural networks. We claim in this paper that the traditional image reconstruction methods and deep learning (DL) are mutually complementary and can be combined to achieve better image reconstruction quality. To test this hypothesis, a hybrid DL image reconstruction method was proposed by combining a state-of-the-art deep learning network, namely a generative adversarial network with cycle loss (CycleGAN), with a traditional data reconstruction algorithm: Projection Onto Convex Set (POCS). The output of the first iteration’s training results of the CycleGAN was updated by POCS and used as the extra training data for the second training iteration of the CycleGAN. The method was validated using sub-sampled Magnetic resonance imaging data. Compared with other state-of-the-art, DL-based methods (e.g., U-Net, GAN, and RefineGAN) and a traditional method (compressed sensing), our method showed the best reconstruction results. Full article

(This article belongs to the Special Issue Machine Learning-Based Medical Image Analysis)

► Show Figures

Figure 1

17 pages, 9340 KiB

Open AccessArticle

Classification of Breast Cancer in Mammograms with Deep Learning Adding a Fifth Class

by Salvador Castro-Tapia, Celina Lizeth Castañeda-Miranda, Carlos Alberto Olvera-Olvera, Héctor A. Guerrero-Osuna, José Manuel Ortiz-Rodriguez, Ma. del Rosario Martínez-Blanco, Germán Díaz-Florez, Jorge Domingo Mendiola-Santibañez and Luis Octavio Solís-Sánchez

Appl. Sci. 2021, 11(23), 11398; https://doi.org/10.3390/app112311398 - 02 Dec 2021

Cited by 7 | Viewed by 2768

Abstract

Breast cancer is one of the diseases of most profound concern, with the most prevalence worldwide, where early detections and diagnoses play the leading role against this disease achieved through imaging techniques such as mammography. Radiologists tend to have a high false positive [...] Read more.

Breast cancer is one of the diseases of most profound concern, with the most prevalence worldwide, where early detections and diagnoses play the leading role against this disease achieved through imaging techniques such as mammography. Radiologists tend to have a high false positive rate for mammography diagnoses and an accuracy of around 82%. Currently, deep learning (DL) techniques have shown promising results in the early detection of breast cancer by generating computer-aided diagnosis (CAD) systems implementing convolutional neural networks (CNNs). This work focuses on applying, evaluating, and comparing the architectures: AlexNet, GoogLeNet, Resnet50, and Vgg19 to classify breast lesions after using transfer learning with fine-tuning and training the CNN with regions extracted from the MIAS and INbreast databases. We analyzed 14 classifiers, involving 4 classes as several researches have done it before, corresponding to benign and malignant microcalcifications and masses, and as our main contribution, we also added a 5th class for the normal tissue of the mammary parenchyma increasing the correct detection; in order to evaluate the architectures with a statistical analysis based on the received operational characteristics (ROC), the area under the curve (AUC), F1 Score, accuracy, precision, sensitivity, and specificity. We generate the best results with the CNN GoogLeNet trained with five classes on a balanced database with an AUC of 99.29%, F1 Score of 91.92%, the accuracy of 91.92%, precision of 92.15%, sensitivity of 91.70%, and specificity of 97.66%, concluding that GoogLeNet is optimal as a classifier in a CAD system to deal with breast cancer. Full article

(This article belongs to the Special Issue Machine Learning-Based Medical Image Analysis)

► Show Figures

Figure 1

20 pages, 6954 KiB

Open AccessArticle

Multiclass Skin Cancer Classification Using Ensemble of Fine-Tuned Deep Learning Models

by Nabeela Kausar, Abdul Hameed, Mohsin Sattar, Ramiza Ashraf, Ali Shariq Imran, Muhammad Zain ul Abidin and Ammara Ali

Appl. Sci. 2021, 11(22), 10593; https://doi.org/10.3390/app112210593 - 11 Nov 2021

Cited by 25 | Viewed by 3893

Abstract

Skin cancer is a widespread disease associated with eight diagnostic classes. The diagnosis of multiple types of skin cancer is a challenging task for dermatologists due to the similarity of skin cancer classes in phenotype. The average accuracy of multiclass skin cancer diagnosis [...] Read more.

Skin cancer is a widespread disease associated with eight diagnostic classes. The diagnosis of multiple types of skin cancer is a challenging task for dermatologists due to the similarity of skin cancer classes in phenotype. The average accuracy of multiclass skin cancer diagnosis is 62% to 80%. Therefore, the classification of skin cancer using machine learning can be beneficial in the diagnosis and treatment of the patients. Several researchers developed skin cancer classification models for binary class but could not extend the research to multiclass classification with better performance ratios. We have developed deep learning-based ensemble classification models for multiclass skin cancer classification. Experimental results proved that the individual deep learners perform better for skin cancer classification, but still the development of ensemble is a meaningful approach since it enhances the classification accuracy. Results show that the accuracy of individual learners of ResNet, InceptionV3, DenseNet, InceptionResNetV2, and VGG-19 are 72%, 91%, 91.4%, 91.7% and 91.8%, respectively. The accuracy of proposed majority voting and weighted majority voting ensemble models are 98% and 98.6%, respectively. The accuracy of proposed ensemble models is higher than the individual deep learners and the dermatologists’ diagnosis accuracy. The proposed ensemble models are compared with the recently developed skin cancer classification approaches. The results show that the proposed ensemble models outperform recently developed multiclass skin cancer classification models. Full article

(This article belongs to the Special Issue Machine Learning-Based Medical Image Analysis)

► Show Figures

Figure 1

17 pages, 4450 KiB

Open AccessArticle

Multi-Class Classification of Lung Diseases Using CNN Models

by Min Hong, Beanbonyka Rim, Hongchang Lee, Hyeonung Jang, Joonho Oh and Seongjun Choi

Appl. Sci. 2021, 11(19), 9289; https://doi.org/10.3390/app11199289 - 06 Oct 2021

Cited by 20 | Viewed by 3990

Abstract

In this study, we propose a multi-class classification method by learning lung disease images with Convolutional Neural Network (CNN). As the image data for learning, the U.S. National Institutes of Health (NIH) dataset divided into Normal, Pneumonia, and Pneumothorax and the Cheonan Soonchunhyang [...] Read more.

In this study, we propose a multi-class classification method by learning lung disease images with Convolutional Neural Network (CNN). As the image data for learning, the U.S. National Institutes of Health (NIH) dataset divided into Normal, Pneumonia, and Pneumothorax and the Cheonan Soonchunhyang University Hospital dataset including Tuberculosis were used. To improve performance, preprocessing was performed with Center Crop while maintaining the aspect ratio of 1:1. As a Noisy Student of EfficientNet B7, fine-tuning learning was performed using the weights learned from ImageNet, and the features of each layer were maximally utilized using the Multi GAP structure. As a result of the experiment, Benchmarks measured with the NIH dataset showed the highest performance among the tested models with an accuracy of 85.32%, and the four-class predictions measured with data from Soonchunhyang University Hospital in Cheonan had an average accuracy of 96.1%, an average sensitivity of 92.2%, an average specificity of 97.4%, and an average inference time of 0.2 s. Full article

(This article belongs to the Special Issue Machine Learning-Based Medical Image Analysis)

► Show Figures

Figure 1

15 pages, 3269 KiB

Open AccessArticle

Image Quality Assessment to Emulate Experts’ Perception in Lumbar MRI Using Machine Learning

by Steren Chabert, Juan Sebastian Castro, Leonardo Muñoz, Pablo Cox, Rodrigo Riveros, Juan Vielma, Gamaliel Huerta, Marvin Querales, Carolina Saavedra, Alejandro Veloz and Rodrigo Salas

Appl. Sci. 2021, 11(14), 6616; https://doi.org/10.3390/app11146616 - 19 Jul 2021

Cited by 10 | Viewed by 2518

Abstract

Medical image quality is crucial to obtaining reliable diagnostics. Most quality controls rely on routine tests using phantoms, which do not reflect closely the reality of images obtained on patients and do not reflect directly the quality perceived by radiologists. The purpose of [...] Read more.

Medical image quality is crucial to obtaining reliable diagnostics. Most quality controls rely on routine tests using phantoms, which do not reflect closely the reality of images obtained on patients and do not reflect directly the quality perceived by radiologists. The purpose of this work is to develop a method that classifies the image quality perceived by radiologists in MR images. The focus was set on lumbar images as they are widely used with different challenges. Three neuroradiologists evaluated the image quality of a dataset that included

T_{1}

-weighting images in axial and sagittal orientation, and sagittal

T_{2}

-weighting. In parallel, we introduced the computational assessment using a wide range of features extracted from the images, then fed them into a classifier system. A total of 95 exams were used, from our local hospital and a public database, and part of the images was manipulated to broaden the distribution quality of the dataset. Good recall of 82% and an area under curve (AUC) of 77% were obtained on average in testing condition, using a Support Vector Machine. Even though the actual implementation still relies on user interaction to extract features, the results are promising with respect to a potential implementation for monitoring image quality online with the acquisition process. Full article

(This article belongs to the Special Issue Machine Learning-Based Medical Image Analysis)

► Show Figures

Graphical abstract

Journal Menu

Journal Browser

Machine Learning-Based Medical Image Analysis

Share This Special Issue

Special Issue Editors

Special Issue Information

Keywords

Published Papers (8 papers)

Research

Further Information

Guidelines

MDPI Initiatives

Follow MDPI