Dataset Information

Different cell imaging methods did not significantly improve immune cell image classification performance.

ABSTRACT: Developments in high-throughput microscopy have made it possible to collect huge amounts of cell image data that are difficult to analyse manually. Machine learning (e.g., deep learning) is often employed to automate the extraction of information from these data, such as cell counting, cell type classification and image segmentation. However, the effects of different imaging methods on the accuracy of image processing have not been examined systematically. We studied the effects of different imaging methods on the performance of machine learning-based cell type classifiers. We observed lymphoid-primed multipotential progenitor (LMPP) and pro-B cells using three imaging methods: differential interference contrast (DIC), phase contrast (Ph) and bright-field (BF). We examined the classification performance of convolutional neural networks (CNNs) with each of them and their combinations. CNNs achieved an area under the receiver operating characteristic (ROC) curve (AUC) of ~0.9, which was significantly better than when the classifier used only cell size or cell contour shape as input. However, no significant differences were found between imaging methods and focal positions.

SUBMITTER: Ogawa T

PROVIDER: S-EPMC8794178 | biostudies-literature | 2022

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Different cell imaging methods did not significantly improve immune cell image classification performance.

Ogawa Taisaku T Ochiai Koji K Iwata Tomoharu T Ikawa Tomokatsu T Tsuzuki Taku T Shiroguchi Katsuyuki K Takahashi Koichi K

PloS one 20220127 1

Developments in high-throughput microscopy have made it possible to collect huge amounts of cell image data that are difficult to analyse manually. Machine learning (e.g., deep learning) is often employed to automate the extraction of information from these data, such as cell counting, cell type classification and image segmentation. However, the effects of different imaging methods on the accuracy of image processing have not been examined systematically. We studied the effects of different ima ...[more]

PMID: 35085287

Similar Datasets

Project description:Chest disease refers to a wide range of conditions affecting the lungs, such as COVID-19, lung cancer (LC), consolidation lung (COL), and many more. When diagnosing chest disorders medical professionals may be thrown off by the overlapping symptoms (such as fever, cough, sore throat, etc.). Additionally, researchers and medical professionals make use of chest X-rays (CXR), cough sounds, and computed tomography (CT) scans to diagnose chest disorders. The present study aims to classify the nine different conditions of chest disorders, including COVID-19, LC, COL, atelectasis (ATE), tuberculosis (TB), pneumothorax (PNEUTH), edema (EDE), pneumonia (PNEU). Thus, we suggested four novel convolutional neural network (CNN) models that train distinct image-level representations for nine different chest disease classifications by extracting features from images. Furthermore, the proposed CNN employed several new approaches such as a max-pooling layer, batch normalization layers (BANL), dropout, rank-based average pooling (RBAP), and multiple-way data generation (MWDG). The scalogram method is utilized to transform the sounds of coughing into a visual representation. Before beginning to train the model that has been developed, the SMOTE approach is used to calibrate the CXR and CT scans as well as the cough sound images (CSI) of nine different chest disorders. The CXR, CT scan, and CSI used for training and evaluating the proposed model come from 24 publicly available benchmark chest illness datasets. The classification performance of the proposed model is compared with that of seven baseline models, namely Vgg-19, ResNet-101, ResNet-50, DenseNet-121, EfficientNetB0, DenseNet-201, and Inception-V3, in addition to state-of-the-art (SOTA) classifiers. The effectiveness of the proposed model is further demonstrated by the results of the ablation experiments. The proposed model was successful in achieving an accuracy of 99.01%, making it superior to both the baseline models and the SOTA classifiers. As a result, the proposed approach is capable of offering significant support to radiologists and other medical professionals.

Project description:OBJECTIVE:To compare the accuracy of different methods for image registration in image-guided adaptive brachytherapy (IGABT) for cervical cancer. METHODS:The last treatment planning CT images (CT1) and the first treatment planning CT images (CT2) were acquired from 15 patients with cervical cancer and registered with different match image qualities (retained/removed catheter source in images) and different match regions [target only (S Group)/ interested organ structure (M Group)/body (L Group)] in Velocity3.2 software. The dice similarity coefficient (DSC) between the clinical target volumes (CTV) of the CT1 and CT2 images (CTVCT1 and CTVCT2, respectively) and between the organs-at-risk (OAR) of the two imaging datasets (OARCT1 and OARCT2, respectively) were used to evaluate the image registration accuracy. RESULTS:The auto-segmentation volume of the catheter source using Velocity software based on the CT threshold was the closest to the actual volume within the CT value range of 1700-1800 HU. In the retained group, the DSC for the OARs of was better than or equal to that of the removed group, and the DSC value of the rectum was significantly improved (P < 0.05). For comparison of different match regions, the high-risk target volume (HRCTV) and the low-risk target volume (IRCTV) had the best precision for registration of the target area, which was significantly greater than that of M group and L group (P < 0.05). The M group had better registration accuracy of the target area and the best accuracy for the OARs. The DSC values of the bladder and rectum were significantly better than those of the other two groups (P < 0.05). CONCLUSIONS:The CT value range of 1700-1800 HU is optimal for automatic image segmentation using Velocity software. Automatic segmentation and shielding the volume of the catheter source can improve the image quality. We recommend the use of interested organ structures regions for image registration in image-guided adaptive brachytherapy for cervical cancer.

Project description:BackgroundThe implementation of deep learning models for medical image classification poses significant challenges, including gradual performance degradation and limited adaptability to new diseases. However, frequent retraining of models is unfeasible and raises concerns about healthcare privacy due to the retention of prior patient data. To address these issues, this study investigated privacy-preserving continual learning methods as an alternative solution.MethodsWe evaluated twelve privacy-preserving non-storage continual learning algorithms based deep learning models for classifying retinal diseases from public optical coherence tomography (OCT) images, in a class-incremental learning scenario. The OCT dataset comprises 108,309 OCT images. Its classes include normal (47.21%), drusen (7.96%), choroidal neovascularization (CNV) (34.35%), and diabetic macular edema (DME) (10.48%). Each class consisted of 250 testing images. For continuous training, the first task involved CNV and normal classes, the second task focused on DME class, and the third task included drusen class. All selected algorithms were further experimented with different training sequence combinations. The final model's average class accuracy was measured. The performance of the joint model obtained through retraining and the original finetune model without continual learning algorithms were compared. Additionally, a publicly available medical dataset for colon cancer detection based on histology slides was selected as a proof of concept, while the CIFAR10 dataset was included as the continual learning benchmark.ResultsAmong the continual learning algorithms, Brain-inspired-replay (BIR) outperformed the others in the continual learning-based classification of retinal diseases from OCT images, achieving an accuracy of 62.00% (95% confidence interval: 59.36-64.64%), with consistent top performance observed in different training sequences. For colon cancer histology classification, Efficient Feature Transformations (EFT) attained the highest accuracy of 66.82% (95% confidence interval: 64.23-69.42%). In comparison, the joint model achieved accuracies of 90.76% and 89.28%, respectively. The finetune model demonstrated catastrophic forgetting in both datasets.ConclusionAlthough the joint retraining model exhibited superior performance, continual learning holds promise in mitigating catastrophic forgetting and facilitating continual model updates while preserving privacy in healthcare deep learning models. Thus, it presents a highly promising solution for the long-term clinical deployment of such models.

Dataset Information

Different cell imaging methods did not significantly improve immune cell image classification performance.

Publications

Different cell imaging methods did not significantly improve immune cell image classification performance.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets