Classification of Breast Tumor from Histopathological Images with Transfer Learning

Histopathological Whole Slide Imaging (WSI) has become a standard in the detection of breast cancer. Automated image analysis methods attempt to reduce the workload from the clinicians and Convolutional Neural Networks (CNNs) are a popular choice for this purpose. However, size of a WSI image typically is approximately <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$40,000\times 40.000$ </tex-math></inline-formula> pixels (can reach up to <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$100,000\times 100.000$ </tex-math></inline-formula> pixels). CNNs cannot handle such large images. Moreover, downscaling a WSI image causes degradation of small-scale visual information. Hence, a large number of small patches (containing critical visual information) from a WSI image are extracted by a trained pathologist and are used for training. However, it requires massive amounts of time to precisely search and label appropriate class-representative patches. To address this issue, a Deep Multiple Instance Learning (MIL) based CNN framework has been introduced in this paper. In the proposed framework every slide is represented as a bag of extracted patches. Only the bag label is used for training, thus eliminating the requirement to provide patchwise labels. The patches inherit the label of the bag containing them. A WSI image (i.e. a bag) is labeled benign if all its patches are benign and labeled malignant even if a single patch contains malignant cells. Learning can be carried out at the bag level even with noisy patch labels. Performance of this method was evaluated using the BreakHis, IUPHL and UCSB breast cancer datasets where 93.06%, 96.63%, 95.83% accuracy was achieved respectively.

Today breast cancer is the leading type of cancer among women undergoing cancer screening. A slight delay in detecting and diagnosing this disease may result in irreversible convolutions. Histopathological images from the biopsy examination present a large amount of structural information that can signi cantly improve the prognosis for breast cancer. The pathological analysis, which involves the microscopic examination of the histopathological slides, is a challenging task. An automated computer-aided detection (CAD) procedure is inevitable, as it may decrease the pathologist examination time and help detect the disease at an early stage. Lately, deep learning methods using arti cial neural networks are consistently in use to improve the performance of CAD methods. A common practice among recent studies is to use the transfer learning approach of training deep neural network architectures. Transfer Learning is an established learning approach that facilitates a deep neural network to train quickly on a speci c dataset and resolve an interdisciplinary problem. Deep Learning methods employing the transfer learning approach have provided highly competitive results on the datasets consisting of the whole slide images, which are captured generally at high resolutions. However, the performance is not remarkably appreciable on the small and low-resolution image datasets, in particular the datasets that include patch samples. In this direction, the present study proposes a novel domain-speci c learning strategy, Breast Histo-Fusion, which aims to detect breast cancer even from images of low resolution and small size. Further, four state-of-the-art deep CNNs (AlexNet, VGG19, ResNet, and DenseNet) are trained using both learning approaches: Transfer Learning and Histo-Fusion on the IDC dataset. The proposed Histo-Fusion learning approach has improved the discriminating abilities and performance of each deep CNN by providing better results of (AlexNet -95.75%, VGG19-95.96%, ResNet34-96.17%, ResNet50-96.67%, and DenseNet121-97.49%) compared to (AlexNet -90.41, VGG19-90.51%, ResNet34-90.83%, ResNet50-92.27%, and DenseNet121-93.05%) using the transfer learning strategy. As a result, the procedure can help expert pathologists to perform accurate diagnoses and reduce false-positive rates.

Presented here are the results of an investigation conducted to determine the effectiveness of deep learning (DL)‐based systems utilizing the power of transfer learning for detecting breast cancer in histopathological images. It is shown that DL models that are not specifically developed for breast cancer detection can be trained using transfer learning to effectively detect breast cancer in histopathological images. The outcome of the analysis enables the selection of the best DL architecture for detecting cancer with high accuracy. This should facilitate pathologists to achieve early diagnoses of breast cancer and administer appropriate treatment to the patient. The experimental work here used the BreaKHis database consisting of 7909 histopathological pictures from 82 clinical breast cancer patients. The strategy presented for DL training uses various image processing techniques for extracting various feature patterns. This is followed by applying transfer learning techniques in the deep convolutional networks like ResNet, ResNeXt, SENet, Dual Path Net, DenseNet, NASNet, and Wide ResNet. Comparison with recent literature shows that ResNext‐50, ResNext‐101, DPN131, DenseNet‐169 and NASNet‐A provide an accuracy of 99.8%, 99.5%, 99.675%, 99.725%, and 99.4%, respectively, and outperform previous studies.

Automated invasive ductal carcinoma detection based using deep transfer learning with whole-slide images

Breast cancer is the prominent foremost disease causing death among the various cancers affecting female. Precise and initial diagnosis of diseases can often save lives. Histopathology microscopy imaging as a best practice used by pathologist to diagnose breast cancer. The procedure is labor intensive and time consuming. Hence, investigation of tissue samples in automated manner from histopathology images has constructive clinical implication in early diagnosis and treatment of disease. To capture discriminating traits from images in better way, images are acquired at various optical magnifications (x40, x100, x200 and x400). The disease can be treated well if the type of actual subclass of benign and malignant categories on a sample image is known early. The research gap observed in current literature isthat, limited study done on magnification independent multi- classification. With recent advances in Artificial Intelligence (AI)especially deep learning, the Convolutional Neural Networks (CNN) can be effectively utilized for automatic histopathology image analysis. In this work, multi-classification framework across different magnification factor using recent tax- onomy of BreakHis dataset is proposed. The proposed approach explores different transfer learning models with fine tuning adaptation. The experiments have shown impressive results on pre-trained models DenseNet121 as well as on DenseNet201. Promising results are observed towards multi-classification of breast cancer histopathology images as compared to other models.

Breast cancer is one of the most common and dangerous cancers in women, while it can also afflict men. Breast cancer treatment and detection are greatly aided by the use of histopathological images since they contain sufficient phenotypic data. A Deep Neural Network (DNN) is commonly employed to improve accuracy and breast cancer detection. In our research, we have analyzed pre-trained deep transfer learning models such as ResNet50, ResNetl01, VGG16, and VGG19 for detecting breast cancer using the 2453 histopathology images dataset. Images in the dataset were separated into two categories: those with invasive ductal carcinoma (IDC) and those without IDC. After analyzing the transfer learning model, we found that ResNet50 outperformed other models, achieving accuracy rates of 90.2%, Area under Curve (AUC) rates of 90.0%, recall rates of 94.7%, and a marginal loss of 3.5%.

The integration of Artificial Intelligence (AI) technologies into healthcare is revolutionising early diagnostic processes for diseases, also for CRC, enhancing both the effectiveness and efficiency of traditional diagnostic methods (4).Fortunately, early detection of CRC increases the 5-year survival rate from 18% when detected at the most advanced stage to 88.5% when detected at an early symptomatic stage (5). Screening programs are implemented for such early detection, as tumours detected in symptomatic patients are larger and at more advanced stages than tumours detected in asymptomatic patients in a screening programme (6). Thus, the detection of colorectal cancer is critical yet challenging, often requiring operatordependant endoscopic procedures, followed by accurate histological assessment of tissue biopsies. AI offers promising enhancements to these traditional methods, providing tools that are less invasive, more accurate, faster, and consistent.In the field of CRC, AI can be applied to different stages of the process, since evaluation of the bowel preparation, before the colonoscopy itself, to histopathology classification, once the lesion has been removed (Figure 1). Following tissue retrieval, Yuan et al. demonstrate an AI algorithm developed through transfer learning from a polyp segmentation model. This localises colorectal cancer regions within precise grids on whole slide imaging (WSI). The method boasts high sensitivity and specificity, and addresses challenges related to AI use and applicability, by accurately labelling cancer presence in histological slides, thus demonstrating reliability and efficacy in histology applications.Regarding to post-colonoscopy stage, Doğan et al. present a study that evaluates the role of artificial intelligence in histopathological image analysis, notably using convolutional neural networks (CNNs) for classifying diverse tissue types within digital pathology. By comparing AI-based algorithms against manual machine learning models, the research demonstrates the superior performance of CNNs in both binary and multi-class classifications of tissues, achieving accuracies of 0.91 and 0.97, respectively. Utilizing over 100,000 images for training and 7,180 for testing, the study highlights AI&#39;s potential to enhance diagnostic accuracy and efficiency in identifying tumour tissues and other types, marking a significant advancement over traditional methods.These studies demonstrate how AI can significantly enhance diagnostic accuracy and efficiency in clinical settings. However, integrating AI tools into healthcare systems will further involve overcoming challenges related to ethical and societal implications, workflow integration and interoperability or interpretability (7).Addressing integration challenges and expanding research to include diverse, multicentre trials will be crucial. This will help validate AI tools&#39; efficacy and safety in real-world settings, ensuring AI tools meet the diverse needs of global populations.The emergence of AI in early CRC diagnosis signifies a significant advance toward more proactive, individualised, and minimally invasive healthcare. Continued research and development are essential to harness the full potential of AI, promising to transform CRC diagnosis and improve patient outcomes and experience.

Breast cancer has evolved as the most lethal illness impacting women all over the globe. Breast cancer may be detected early, which reduces mortality and increases the chances of a full recovery. Researchers all around the world are working on breast cancer screening tools based on medical imaging. Deep learning approaches have piqued the attention of many in the medical imaging field due to their rapid growth. In this research, mammography pictures were utilized to detect breast cancer. We have used four mammography imaging datasets with a similar number of 1145 normal, benign, and malignant pictures using various deep CNN (Inception V4, ResNet-164, VGG-11, and DenseNet121) models as base classifiers. The proposed technique employs an ensemble approach in which the Gompertz function is used to build fuzzy rankings of the base classification techniques, and the decision scores of the base models are adaptively combined to construct final predictions. The proposed fuzzy ensemble techniques outperform each individual transfer learning methodology as well as multiple advanced ensemble strategies (Weighted Average, Sugeno Integral) with reference to prediction and accuracy. The suggested Inception V4 ensemble model with fuzzy rank based Gompertz function has a 99.32% accuracy rate. We believe that the suggested approach will be of tremendous value to healthcare practitioners in identifying breast cancer patients early on, perhaps leading to an immediate diagnosis.

Simple SummaryBreast cancer is one of the leading causes of cancer death among women. Developing machine learning-based diagnosis models receives great attention from researchers and scientists using histopathology images. Deep learning (DL) algorithms automatically extract features from raw data through convolutional operations. The generalization of the DL models’ results relies on large datasets, although they eliminate the expert knowledge in the feature extraction stage. This work aimed to compare the performance of the features extracted via deep learning and a knowledge-based approach in breast cancer detection from histopathology images.Cancer develops when a single or a group of cells grows and spreads uncontrollably. Histopathology images are used in cancer diagnosis since they show tissue and cell structures under a microscope. Knowledge-based and deep learning-based computer-aided detection is an ongoing research field in cancer diagnosis using histopathology images. Feature extraction is vital in both approaches since the feature set is fed to a classifier and determines the performance. This paper evaluates three feature extraction methods and their performance in breast cancer diagnosis. Features are extracted by (1) a Convolutional Neural Network, (2) a transfer learning architecture VGG16, and (3) a knowledge-based system. The feature sets are tested by seven classifiers, including Neural Network (64 units), Random Forest, Multilayer Perceptron, Decision Tree, Support Vector Machines, K-Nearest Neighbors, and Narrow Neural Network (10 units) on the BreakHis 400× image dataset. The CNN achieved up to 85% for the Neural Network and Random Forest, the VGG16 method achieved up to 86% for the Neural Network, and the knowledge-based features achieved up to 98% for Neural Network, Random Forest, Multilayer Perceptron classifiers.

In the realm of computer vision, image classification is a critical issue with many applications, including multimedia content analysis, security and surveillance, and medical imaging. The accuracy of image classification algorithms has considerably increased with the development of deep learning. The discipline of image classification has dramatically benefited from the usage of pre-trained Convolutional Neural Network (CNN) models. In this study, we perform image classification on the Wang dataset composed of 1000 images separated into ten categories, using six different pre-trained CNN models. The research made use of pre-trained versions of VGG16, Densenet, Mobilenet, Inception V3, Resnet50, and Xception models. We assessed the model performances in terms of training and testing accuracy. Testing accuracies for ten unique batches of data were calculated and averaged out. Densenet outperformed state-of-the-art models like Xception by a small margin, mainly due to low data availability. The findings of this study show how pre-trained models have advanced the field of image classification and shed light on how well these models perform in diverse image classification tasks. This study can serve as a valuable reference for researchers and practitioners in the field of computer vision and deep learning and help inform their choices when selecting pre-trained models for image classification, depending on their needs, while also considering data availability and computational constraints.

Diabetes affects 40–45% of Diabetic Retinopathy (DR) patients in the US. Early detection of DR may prevent or postpone vision deterioration, but it is difficult since the disorder often manifests with few symptoms until it is too late to treat. Clinically, DR is routinely treated using fundus images, with an estimated 200 million cases worldwide and over 400,000 deaths each year. A great deal of progress has been made by applying machine learning algorithms to the fundus images. As a result, image classification and detection have become reliable techniques for detecting the severity of diabetic retinopathy. Convolutional Neural Networks (CNNs) play a crucial role in the image classification and detection process by capturing various images' details, enabling a fast and efficient method for detecting diabetic retinopathy. CNN pre-trained models such as ResNet50, Inception V3,and EfficientNetB7 have substantially improved their performance in the ImageNet Large-Scale Visual Recognition Competition. In addition, these pre-trained models are more precise and inexpensive to train because of the shorter connections between their input and output layers. This work proposes an approach for image classification that ensembles three pre-trained models, namely: EfficientNetB7, ResNet50V2, and Inception V3,to perform the classification of the diabetic retinopathy subtypes. Our proposed method achieves 97.43 % accuracy by adjusting the weights of the pre-trained models in detecting D R using the EyePacs Dataset.

Pathologists are required to integrate data from multiple sources when making a diagnosis. Furthermore, whole slide imaging (WSI) and next generation sequencing will escalate data size and complexity. Development of well-designed databases that can allow efficient navigation between multiple data types is necessary for both clinical and research purposes. We developed and evaluated an interactive, web-based database that integrates clinical, histologic, immunohistochemical and genetic information to aid in pathologic diagnosis and interpretation with nine lung adenocarcinoma cases. To minimize sectioning artifacts, representative blocks were serially sectioned using automated tissue sectioning (Kurabo Industries, Osaka Japan) and selected slides were stained by multiple techniques, (hematoxylin and eosin [H&E], immunohistochemistry [IHC] or fluorescence in situ hybridization [FISH]). Slides were digitized by WSI scanners. An interactive relational database was designed based on a list of proposed fields covering a variety of clinical, pathologic and molecular parameters. By focusing on the three main tasks of 1.) efficient management of textual information, 2.) effective viewing of all varieties of stained whole slide images (WSI), and 3.) assistance in evaluating WSI with computer-aided diagnosis, this database prototype shows great promise for multi-modality research and diagnosis.

We read with interest the recently published pilot study on digital whole slide imaging (WSI) for cervicovaginal cytology by Wright and colleagues.1 Their study employed 7 cytologists to review 11 Papanicolaou test slides using digital and conventional light microscope modalities. We acknowledge that their data indicate greater diagnostic accuracy and shorter mean times to reach diagnoses with Papanicolaou test glass slides. However, we would like to express our concern regarding the design of that study and the subsequent conclusion drawn, based on those data, that WSI is not yet ready for routine cytology use.Although we recognize that this was a pilot study, in our opinion, 11 cases is insufficient to reliably measure discordance between digital and conventional microscopy nor is it enough to ensure that users are familiar with a new technology. Prior validation studies comparing WSI and glass slides have demonstrated that, when too few cases are used, the diagnostic accuracy deteriorates. A meta-analysis of the literature indicates that, when an average of 20 cases was employed, the concordance between WSI and glass slides was 75% versus 95% when 60 cases were used.2 Furthermore, although the authors do not specify whether any WSI training was provided in their study, it is unlikely that those cytologists received equivalent exposure to digitized slides as they have had in handling glass slides during their careers. Indeed, previous published data show that accuracy with WSI was 79% without training pathologists to use that technology, compared with 95% when they did receive training.2 The authors do comment that their disagreement rates may, in part, be explained by their lack of familiarity with this new technology.The authors further remark about the current lack of available commercial technology to scan the edges of a glass slide, to help methodically screen WSI slides, and the inability of systems to handle large digital files. We would like to state that there are WSI scanners that are capable of scanning the entire glass slide (Figure), even including the label if required. Their study does highlight an important point, however, ensuring that all material present on a glass slide to be scanned is included in the digital image. Although we admit that it can be challenging to screen a digitized cytology slide, we also feel that it is important to point out that the task can be made much easier using currently available WSI viewer tools for easy and systematic navigation of virtual slides (such as thumbnails, autopanning, keyboard navigation, and tracker capabilities).3 It is important to train users to make use of these tools when viewing virtual slides. It is true that scanning a glass slide generates large WSI digital files. That is why vendors of those WSI devices specify that users employ adequate computers and monitors to handle and view the digital images. The computer workstation (1 gigabyte, random access memory, 64 video random access memory [1 GB RAM, 64 VRAM]) and display monitor (1600 by 1024 pixel resolution) the authors selected for their study seems to have less than the minimum required specifications for the iScan Coreo (Ventana Medical Systems, Tucson, Arizona) instrument they used (2 GB RAM and 1920 by 1200 pixel resolution). That may explain why their system frequently “froze,” causing delay and frustration for their participants. For digital imaging, the workstation's specifications should to be at least, if not higher, than those recommended by the vendor to achieve proper performance and user satisfaction.In conclusion, although we agree with Wright et al that there are still several obstacles to overcome, we believe that, if used effectively, WSI will soon be adopted for daily cytology diagnostic use. To accomplish that, laboratories will need to first invest in the necessary technology, appropriately validate those systems, and adequately train their cytologists to use them.

Commentary: Can pathologists interpret digital images as well as they interpret microscope slides?

Context:In recent years, a new concept has emerged at the forefront of slide-based diagnosis and telepathology. This is the concept of whole slide imaging. This has very recently also been tried with smartphones using complex software for photo analysis and stitching. This study is aimed to evaluate and compare the use of Jenoptik photomicrograph camera and smartphone camera-based whole slide imaging (WSI) for various microscopic preparations and slides. The study also uses a commonly available IOS smartphone device, a commonly available phone to microscope attachment, and the most commonly used Adobe Photoshop software for all stitching purposes.Aim:To evaluate and compare smartphone and photomicrography-based whole slide imaging.Subjects and Methods:The study was conducted at the Department of Oral Pathology and Microbiology in Bapuji Dental College and Hospital, Davangere. The samples included in this study are 10 slides of routine hematoxylin and eosin stain slides and 10 ground section slides of teeth and bone samples. Continuous photographs of the entire slide were captured using an IOS device and a Jenoptik camera attached research microscope at 10× magnification. The photographs were stitched using Adobe Photoshop 2017 software to obtain a whole slide image. Then, two observers analyzed the whole slide image for a possible diagnosis and thus compared the efficiency of both the methods.Statistical Analysis Used:The statistical Chi-square test and analysis of variance (ANOVA) H test were done using Statistical Package for the Social Sciences (SPSS) 2010 Software.Results:Percentage of measure of agreement was 79.5%, 83.7% and 86.3%, 89.1% for the whole slide images taken in the smartphone and photomicrograph and analyzed by Observer 1 and Observer 2, respectively.Conclusions:In this study, it can be seen that the WSI by research microscopes is better than that with a smartphone and a compound microscope but still both the methods are equally good and can be followed with accurate results.