RESUMO
BACKGROUND: A major hurdle for the real time deployment of the AI models is ensuring trustworthiness of these models for the unseen population. More often than not, these complex models are black boxes in which promising results are generated. However, when scrutinized, these models begin to reveal implicit biases during the decision making, particularly for the minority subgroups. METHOD: We develop an efficient adversarial de-biasing approach with partial learning by incorporating the existing concept activation vectors (CAV) methodology, to reduce racial disparities while preserving the performance of the targeted task. CAV is originally a model interpretability technique which we adopted to identify convolution layers responsible for learning race and only fine-tune up to that layer instead of fine-tuning the complete network, limiting the drop in performance RESULTS:: The methodology has been evaluated on two independent medical image case-studies - chest X-ray and mammograms, and we also performed external validation on a different racial population. On the external datasets for the chest X-ray use-case, debiased models (averaged AUC 0.87 ) outperformed the baseline convolution models (averaged AUC 0.57 ) as well as the models trained with the popular fine-tuning strategy (averaged AUC 0.81). Moreover, the mammogram models is debiased using a single dataset (white, black and Asian) and improved the performance on an external datasets (averaged AUC 0.8 to 0.86 ) with completely different population (primarily Hispanic patients). CONCLUSION: In this study, we demonstrated that the adversarial models trained only with internal data performed equally or often outperformed the standard fine-tuning strategy with data from an external setting. The adversarial training approach described can be applied regardless of predictor's model architecture, as long as the convolution model is trained using a gradient-based method. We release the training code with academic open-source license - https://github.com/ramon349/JBI2023_TCAV_debiasing.
Assuntos
Inteligência Artificial , Tomada de Decisão Clínica , Diagnóstico por Imagem , Grupos Raciais , Humanos , Mamografia , Grupos Minoritários , Viés , Disparidades em Assistência à SaúdeRESUMO
BACKGROUND OF THE STUDY: Breast cancer is the most fatal disease that widely affects women. When the cancerous lumps grow from the cells of the breast, it causes breast cancer. Self-analysis and regular medical check-ups help for detecting the disease earlier and enhance the survival rate. Hence, an automated breast cancer detection system in mammograms can assist clinicians in the patient's treatment. In medical techniques, the categorization of breast cancer becomes challenging for investigators and researchers. The advancement in deep learning approaches has established more attention to their advantages to medical imaging issues, especially for breast cancer detection. AIM: The research work plans to develop a novel hybrid model for breast cancer diagnosis with the support of optimized deep-learning architecture. METHODS: The required images are gathered from the benchmark datasets. These collected datasets are used in three pre-processing approaches like "Median Filtering, Histogram Equalization, and morphological operation", which helps to remove unwanted regions from the images. Then, the pre-processed images are applied to the Optimized U-net-based tumor segmentation phase for obtaining accurate segmented results along with the optimization of certain parameters in U-Net by employing "Adapted-Black Widow Optimization (A-BWO)". Further, the detection is performed in two different ways that is given as model 1 and model 2. In model 1, the segmented tumors are used to extract the significant patterns with the help of the "Gray-Level Co-occurrence Matrix (GLCM) and Local Gradient pattern (LGP)". Further, these extracted patterns are utilized in the "Dual Model accessed Optimized Long Short-Term Memory (DM-OLSTM)" for performing breast cancer detection and the detected score 1 is obtained. In model 2, the same segmented tumors are given into the different variants of CNN, such as "VGG19, Resnet150, and Inception". The extracted deep features from three CNN-based approaches are fused to form a single set of deep features. These fused deep features are inserted into the developed DM-OLSTM for getting the detected score 2 for breast cancer diagnosis. In the final phase of the hybrid model, the score 1 and score 2 obtained from model 1 and model 2 are averaged to get the final detection output. RESULTS: The accuracy and F1-score of the offered DM-OLSTM model are achieved at 96 % and 95 %. CONCLUSION: Experimental analysis proves that the recommended methodology achieves better performance by analyzing with the benchmark dataset. Hence, the designed model is helpful for detecting breast cancer in real-time applications.
Assuntos
Mamografia , Neoplasias , Feminino , AnimaisRESUMO
One of the most frequent cancers in women is breast cancer, and in the year 2022, approximately 287,850 new cases have been diagnosed. From them, 43,250 women died from this cancer. An early diagnosis of this cancer can help to overcome the mortality rate. However, the manual diagnosis of this cancer using mammogram images is not an easy process and always requires an expert person. Several AI-based techniques have been suggested in the literature. However, still, they are facing several challenges, such as similarities between cancer and non-cancer regions, irrelevant feature extraction, and weak training models. In this work, we proposed a new automated computerized framework for breast cancer classification. The proposed framework improves the contrast using a novel enhancement technique called haze-reduced local-global. The enhanced images are later employed for the dataset augmentation. This step aimed at increasing the diversity of the dataset and improving the training capability of the selected deep learning model. After that, a pre-trained model named EfficientNet-b0 was employed and fine-tuned to add a few new layers. The fine-tuned model was trained separately on original and enhanced images using deep transfer learning concepts with static hyperparameters' initialization. Deep features were extracted from the average pooling layer in the next step and fused using a new serial-based approach. The fused features were later optimized using a feature selection algorithm known as Equilibrium-Jaya controlled Regula Falsi. The Regula Falsi was employed as a termination function in this algorithm. The selected features were finally classified using several machine learning classifiers. The experimental process was conducted on two publicly available datasets-CBIS-DDSM and INbreast. For these datasets, the achieved average accuracy is 95.4% and 99.7%. A comparison with state-of-the-art (SOTA) technology shows that the obtained proposed framework improved the accuracy. Moreover, the confidence interval-based analysis shows consistent results of the proposed framework.
RESUMO
Breast cancer has become a common malignancy in women. However, early detection and identification of this disease can save many lives. As computer-aided detection helps radiologists in detecting abnormalities efficiently, researchers across the world are striving to develop reliable models to deal with. One of the common approaches to identifying breast cancer is through breast mammograms. However, the identification of malignant breasts from mass lesions is a challenging research problem. In the current work, we propose a method for the classification of breast mass using mammograms which consists of two main stages. At first, we extract deep features from the input mammograms using the well-known VGG16 model while incorporating an attention mechanism into this model. Next, we apply a meta-heuristic called Social Ski-Driver (SSD) algorithm embedded with Adaptive Beta Hill Climbing based local search to obtain an optimal features subset. The optimal features subset is fed to the K-nearest neighbors (KNN) classifier for the classification. The proposed model is demonstrated to be very useful for identifying and differentiating malignant and healthy breasts successfully. For experimentation, we evaluate our model on the digital database for screening mammography (DDSM) database and achieve 96.07% accuracy using only 25% of features extracted by the attention-aided VGG16 model. The Python code of our research work is publicly available at: https://github.com/Ppayel/BreastLocalSearchSSD.
RESUMO
Traditional breast cancer detection algorithms require manual extraction of features from mammogram images and professional medical knowledge. Still, the quality of mammogram images hampers this and extracting high-quality features, which can result in very long processing times. Therefore, this paper proposes a new K-means++ clustering based on Cuckoo Search Optimization (KM++CSO) for breast cancer detection. The pre-processing method is used to improve the proposed KM++CSO method more segmentation efficiently. Furthermore, the interpretability is further enhanced using mathematical morphology and OTSU's threshold. To this end, we tested the effectiveness of the KM++CSO methods on the mammogram image analysis society of the Mini-Mammographic Image Analysis Society (Mini-MIAS), the Digital Database for Screening Mammography (DDSM), and the Breast Cancer Digital Repository (BCDR) dataset through cross-validation. We maximize the accuracy and Jaccard index score, which is a measure that indicates the similarity between detected cancer and their corresponding reference cancer regions. The experimental results showed that the detection method obtained an accuracy of 96.42% (Mini-MIAS), 95.49% (DDSM), and 96.92% (BCDR). On overage, the KM++CSO method obtained 96.27% accuracy for three publicly available datasets. In addition, the detection results provided the 91.05% Jaccard index score.
RESUMO
Breast cancer is widespread worldwide and can be cured if diagnosed early. Using digital mammogram images and image processing with artificial intelligence can play an essential role in breast cancer diagnosis. As many computerized algorithms for breast cancer diagnosis have significant limitations, such as noise handling and varying or low contrast in the images, it can be difficult to segment the abnormal region. These challenges could be overcome by proposing a new pre-processing model, exploring its impact on the post-processing module, and testing it on an extensive database. In this research work, the three-step method is proposed and validated on large databases of mammography images. The first step corresponded to the database classification, followed by the second step, which removed the pectoral muscle from the mammogram image. The third stage utilized new image-enhancement techniques and a new segmentation module to detect abnormal regions in a well-enhanced image to diagnose breast cancer. The pre-and post-processing modules are based on novel image processing techniques. The proposed method was tested using data collected from different hospitals in the Qassim Health Cluster, Qassim Province, Saudi Arabia. This database contained the five categories in the Breast Imaging and Reporting and Data System and consisted of 2892 images; the proposed method is analyzed using the publicly available Mammographic Image Analysis Society database, which contained 322 images. The proposed method gives good contrast enhancement with peak-signal to noise ratio improvement of 3 dB. The proposed method provides an accuracy of approximately 92% on 2892 images of Qassim Health Cluster, Qassim Province, Saudi Arabia. The proposed method gives approximately 97% on the Mammographic Image Analysis Society database. The novelty of the proposed work is that it could work on all Breast Imaging and Reporting and Data System categories. The performance of the proposed method demonstrated its ability to improve the diagnostic performance of the computerized breast cancer detection method.
RESUMO
BACKGROUND: Breast cancer is one of the most common cancers in women. Mammogram images have an important role in the treatment of various states of this cancer. In recent years, machine learning methods have been widely used for tumor segmentation in mammogram images. Pixel-based segmentation methods have been presented using both supervised and unsupervised learning approaches. Supervised learning methods are usually fast and accurate, but they usually use a large number of labeled data. Besides, providing these samples is very hard and usually expensive. Unsupervised learning methods do not require the labels of the training data for decision making and they completely ignore the prior knowledge that may lead to a low performance. Semi-supervised learning methods which use a small number of labeled data solve the problem of providing the high number of samples in supervised methods, while they usually result in a higher accuracy in comparison to the unsupervised methods. METHODS: In this study, we used a semisupervised method for tumor segmentation in which the pixel information is used for the classification. The static and gray level run length matrix features for each pixel are considered as the features, and Fisher discriminant analysis (FDA) is used for feature reduction. A cotraining algorithm based on support vector machine and Bayes classifiers is proposed for tumor segmentation on MIAS data set. RESULTS AND CONCLUSION: The results show that the proposed method outperforms both supervised methods.
RESUMO
We evaluated whether using synthetic mammograms for training data augmentation may reduce the effects of overfitting and increase the performance of a deep learning algorithm for breast mass detection. Synthetic mammograms were generated using in silico procedural analytic breast and breast mass modeling algorithms followed by simulated x-ray projections of the breast models into mammographic images. In silico breast phantoms containing masses were modeled across the four BI-RADS breast density categories, and the masses were modeled with different sizes, shapes, and margins. A Monte Carlo-based x-ray transport simulation code, MC-GPU, was used to project the three-dimensional phantoms into realistic synthetic mammograms. 2000 mammograms with 2522 masses were generated to augment a real data set during training. From the Curated Breast Imaging Subset of the Digital Database for Screening Mammography (CBIS-DDSM) data set, we used 1111 mammograms (1198 masses) for training, 120 mammograms (120 masses) for validation, and 361 mammograms (378 masses) for testing. We used faster R-CNN for our deep learning network with pretraining from ImageNet using the Resnet-101 architecture. We compared the detection performance when the network was trained using different percentages of the real CBIS-DDSM training set (100%, 50%, and 25%), and when these subsets of the training set were augmented with 250, 500, 1000, and 2000 synthetic mammograms. Free-response receiver operating characteristic (FROC) analysis was performed to compare performance with and without the synthetic mammograms. We generally observed an improved test FROC curve when training with the synthetic images compared to training without them, and the amount of improvement depended on the number of real and synthetic images used in training. Our study shows that enlarging the training data with synthetic samples can increase the performance of deep learning systems.
RESUMO
Mammogram images are majorly used for detecting the breast cancer. The level of positivity of breast cancer is detected after excluding the pectoral muscle from mammogram images. Hence, it is very significant to identify and segment the pectoral muscle from the mammographic images. In this work, a new multilevel thresholding, on the basis of electro-magnetism optimization (EMO) technique, is proposed. The EMO works on the principle of attractive and repulsive forces among the charges to develop the members of a population. Here, both Kapur's and Otsu based cost functions are employed with EMO separately. These standard functions are executed over the EMO operator till the best solution is achieved. Thus, optimal threshold levels can be identified for the considered mammographic image. The proposed methodology is applied on all the three twenty-two mammogram images available in mammographic image analysis society dataset, and successful segmentation of the pectoral muscle is achieved for majority of the mammogram images. Hence, the proposed algorithm is found to be robust for variations in the pectoral muscle.
RESUMO
Objective: Generally, medical images contain lots of noise that may lead to uncertainty in diagnosing the abnormalities. Computer aided diagnosis systems offer a support to the radiologists in identifying the disease affected area. In mammographic images, some normal tissues may appear to be similar to masses and it is tedious to differentiate them. Therefore, this paper presents a novel framework for the detection of mammographic masses that leads to early diagnosis of breast cancer. Methods: This work proposes a Crow search optimization based Intuitionistic fuzzy clustering approach with neighborhood attraction (CrSA-IFCM-NA) for identifying the region of interest. First order moments were extracted from preprocessed images. These features were given as input to the Intuitionistic fuzzy clustering algorithm. Instead of randomly selecting the initial centroids, crow search optimization technique is applied to choose the best initial centroid and the masses are separated. Experiments are conducted over the images taken from the Mammographic Image Analysis Society (mini-MIAS) database. Results: CrSA-IFCM-NA effectively separated the masses from mammogram images and proved to have good results in terms of cluster validity indices indicating the clear segmentation of the regions. Conclusion: The experimental results show that the accuracy of the proposed method proves to be encouraging for detection of masses. Thus, it provides a better assistance to the radiologists in diagnosing breast cancer at an early stage.
Assuntos
Neoplasias da Mama/diagnóstico , Algoritmos , Análise por Conglomerados , Lógica Fuzzy , Humanos , Processamento de Imagem Assistida por Computador/métodos , Mamografia/métodosRESUMO
A new algorithm which uses tree based decision and tri-state non linear values to eliminate high density outlier noise in mammogram images is proposed. The proposed algorithm uses the number of non noisy pixels in the current processing vicinity as values in the decision tree. Tri-state values such as unsymmetrical truncated median or modified Winsorized mean or midpoint replaces the corrupted pixel based on the decision tree in the current processing kernel. The algorithm exhibits good PSNR, IEF, low MSE and High structural preservation property even after removing high density noise. The performance of the proposed algorithm was also found good visually. The Key aspect of the work is the combination of tree based decision and tri-state non linear values which preserves the information content of images that are required for further processing.
Assuntos
Algoritmos , Processamento de Imagem Assistida por Computador/métodos , Mamografia/métodos , Simulação por Computador , Árvores de Decisões , Humanos , Intensificação de Imagem Radiográfica/métodos , Razão Sinal-RuídoRESUMO
Purpose: Breast cancer can be cured if diagnosed early, with digital mammography which is one of the most effective imaging modalities for early detection. However mammogram images often come with low contrast, high background noises and artifacts, making diagnosis difficult. The purpose of this research is to preprocess mammogram images to improve results with a computer aided diagnosis system. The focus is on three preprocessing methods: a breast border segmentation method; a contrast enhancement method; and a pectoral muscle removal method. Methods: The proposed breast border extraction method employs a threshold based segmentation technique along with a combination of morphological operations. The contrast enhancement method presented here is divided into two phages. In phase I, a bi-level histogram modification technique is applied to enhance the image globally and in phase II a non-linear filter based on local mean and local standard deviation for each pixel is applied to the histogram modified image. The pectoral muscle removal method discussed here is implemented by applying a region growing algorithm. Results: The proposed techniques are tested with the Mini MIAS dataset. The breast border extraction method is applied to 322 images and achieved 98.7% segmentation accuracy. The contrast enhancement method is evaluated based on quantitative measures like measure of enhancement, absolute mean brightness error, combined enhancement measure and discrete entropy. The proposed contrast enhancement method when applied to 14 images with different types of masses, the quantitative measures showed an optimum level of contrast enhancement compared to other enhancement methods with preservation of local detail. Removal of the pectoral muscle from MLO mammogram images reduced the search region while identifying abnormalities like masses and calcification. Conclusions: The preprocessing steps proposed here show promising results in terms of both qualitative and quantitative analysis.