A STUDY ON BRAIN TUMOR TECHNOLOGY USING MACHINE LEARNING WITH FEWER IMAGES ON TRAINING WITH HIGH ACCURACY

A STUDY ON BRAIN TUMOR TECHNOLOGY USING MACHINE LEARNING WITH FEWER IMAGES ON TRAINING WITH HIGH ACCURACY | Authorea try { document.documentElement.classList.add('js'); } catch (e) { } var _gaq = _gaq || []; _gaq.push(['_setAccount', 'G-8VDV14Y67G']); _gaq.push(['_trackPageview']); (function() { var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true; ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js'; var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s); })(); Skip to main content Preprints Collections Wiley Open Research IET Open Research Ecological Society of Japan All Collections About About Authorea FAQs Contact Us Quick Search anywhere Search for preprint articles, keywords, etc. Search Search ADVANCED SEARCH SCROLL This is a preprint and has not been peer reviewed. Data may be preliminary. 3 December 2025 V1 Latest version Share on A STUDY ON BRAIN TUMOR TECHNOLOGY USING MACHINE LEARNING WITH FEWER IMAGES ON TRAINING WITH HIGH ACCURACY Author : Kandadi Thirupathi Reddy 0009-0007-0835-0581 [email protected] Authors Info & Affiliations https://doi.org/10.22541/au.176478088.86005575/v1 260 views 167 downloads Contents Abstract Literature Review: Results and Discussions Results: Conclusion References Information & Authors Metrics & Citations View Options References Figures Tables Media Share Abstract In medical practice, brain tumours are still difficult to identify early and diagnose accurately, which frequently results in postponed therapies and unfavourable patient outcomes. Promising approaches to improve brain tumour diagnosis, classification, and prognosis are provided by recent developments in machine learning (ML) technologies. With an emphasis on imaging data like MRI and CT scans, this study explores the use of machine learning approaches in brain tumour detection and management. The effectiveness of several machine learning methods, such as random forests, support vector machines, and deep learning, in automating tumour recognition, segmentation, and classification is investigated. Through the analysis of imaging features and clinical data, the study also assesses the ability of machine learning algorithms to forecast tumour growth and patient survival outcomes. Initial findings show that ML models can perform better than conventional. Keywords: Image segmentation, CNN, Augmentation, Image classification, MRI Introduction Because of their intricacy, unpredictable nature, and the brain’s crucial position inside the body, brain tumours provide a serious medical problem. Improving patient outcomes requires early detection, precise diagnosis, and efficient treatment of brain tumours. Although they can be time-consuming and frequently require expert interpretation, traditional diagnostic techniques including MRI scans, CT scans, and histopathology examination have been employed extensively. Machine learning (ML) has become a potent instrument for enhancing medical diagnostics in recent years, especially in the area of brain tumour identification and categorisation. Machine learning models can help identify trends and anomalies that human specialists would overlook by using sophisticated algorithms that can learn from large datasets. These technologies have the potential to enhance the Ultimately, this research aims to contribute to the development of more efficient, reliable, and accessible tools for the diagnosis and treatment of brain tumors, improving patient care and clinicaloutcomes. FIG 1: A sample of MRI images from the brain tumor dataset Literature Review: One important and difficult topic in medical imaging and diagnostics is brain tumors. In order to improve patient outcomes, researchers have worked to improve the early identification, diagnosis, and prognosis of brain tumors using advances in machine learning (ML). The present status of research on applying machine learning approaches for brain tumor detection, classification, and prediction from several imaging modalities is examined in this overview of the literature. 1.Importance of Early Detection of Brain Tumors There are two types of brain tumours: malignant and benign. Because of their propensity to spread to neighbouring tissues and their quick growth, malignant tumours are especially deadly. A successful course of treatment depends on an early diagnosis. Imaging methods that provide comprehensive structural information, such as Computed Tomography (CT) scans and Magnetic Resonance Imaging (MRI), have historically been used to detect brain tumours. However, manually analysing these photos takes a lot of time, is prone to mistakes, and requires a lot of knowledge. 2. Machine Learning in Medical Imaging Over the past few years, medical imaging has advanced significantly thanks to machine learning (ML), which offers solutions for tasks like segmentation, classification, and prediction. Machine learning algorithms can help detect brain tumours, categorise them into several categories (such as gliomas, meningiomas, and pituitary tumours), and forecast patient outcomes. In brain tumour research, common machine learning techniques include: Supervised Learning: Using labelled training data, algorithms like Support Vector Machines (SVM), Random Forests (RF), and deep learning models are used to classify tumour images. Unsupervised Learning: Without first labelling the tumour locations, clustering techniques such as k-means are utilised to identify abnormalities in brain scans. Reinforcement Learning: Although less popular, techniques for reinforcement learning are being investigated for ongoing enhancements in the precision of tumour diagnosis. 3. Deep Learning Models in Brain Tumor Detection Because it can learn complex characteristics from data without the need for manual feature extraction, deep learning—a subset of machine learning—has emerged as the most popular method in recent years. The ability of Convolutional Neural Networks (CNNs) to identify spatial hierarchies in image data has made them very effective in the analysis of medical images. CNNs have been used in numerous studies to automatically detect and segment tumours in MRI and CT scans,with impressive results. For instance, the VGGNet and ResNet designs have demonstrated remarkable efficacy in distinguishing between different types of brain tumours when applied to the classification of brain tumour images. Furthermore, the U-Net design has become more and more popular for segmentation jobs since it makes it possible to precisely locate tumour regions. 4. Data Preprocessing and Augmentation Effective preprocessing is crucial for improving the quality of input data since brain tumour images are complicated and vary widely. To increase the accuracy of machine learning models, methods including noise reduction, contrast enhancement, and normalisation are frequently employed. Moreover, data augmentation methods like rotation, flipping, and scaling are used to artificially increase the training dataset and enhance model generalisation because labelled datasets are scarce. 5. Challenges in Brain Tumor Detection There are still a number of obstacles to overcome despite the encouraging developments in machine learning for brain tumour detection: Data Imbalance: A higher percentage of normal brain photos than tumour images is a common feature of tumour datasets. Model bias may result from this, making it possible for the algorithm to miss uncommon tumour types. Inter- and Intra-patient Variability: Because of variables such tumour size, location, and patient-specific traits, brain tumour pictures can differ greatly between patients. One of the biggest challenges is creating reliable models that generalise well across these variances. Interpretability of Models: Although deep learning models have a high degree of accuracy, they frequently operate as ”black boxes,” making it challenging for medical professionals to comprehend the decision-making process. In clinical contexts, this lack of interpretability may restrict adoption and confidence. 6. Recent Research and Applications Numerous research have shown how machine learning can be used to detect and categorise brain tumours: Bakas et al. (2017) achieved state-of-the-art results in the BRATS (Brain Tumour Segmentation) challenge by proposing a deep learning model for brain tumour segmentation on MRI data. Cai et al. (2020) investigated the automatic categorisation of brain tumours from MRI scans using a hybrid model that combines CNNs and recurrent neural networks (RNNs). Their model outperformed conventional techniques in terms of accuracy. In order to increase the accuracy of tumour classification and patient prognosis prediction, Gibson et al. (2022) concentrated on integrating multimodal data (such as MRI, CT, and genetic data) utilising ensemble learning techniques. 7. Future Directions The detection of brain tumours using machine learning is still developing. Future research in a number of prospective areas includes: Multimodal Imaging: Combining information from several imaging modalities (such as MRI, CT, and PET scans) might offer a more thorough understanding of brain tumours and possibly increase the precision of diagnosis. Transfer Learning: This technique can assist address data scarcity concerns, especially in specialised medical datasets, by applying previously trained models from one domain to brain tumour detection tasks. Explainable AI (XAI): More attention is being paid to creating models that offer insights into decision-making processes in addition to precise forecasts. As a result, ML models will be more widely accepted and trusted in therapeutic contexts. Personalised Medicine: By improving machine learning models, individual patient predictions can be made. PROPOSED SYSTEM The accuracy and speed of detecting brain tumours using medical pictures (such as MRI and CT scans) could be greatly increased by a machine learning-based brain tumour detection and diagnosis system. This system would handle and analyse medical data using sophisticated machine learning algorithms, especially deep learning. An outline of the possible structure of such a system is as follows: 1. Data Collection and Preprocessing Medical Imaging Data: Compile a sizable collection of brain scans (CT, MRI) that have been labelled with tumour classifications (e.g., tumour type, benign or malignant). Preprocessing of Data: Normalisation: Make the pixel/voxel data’s intensity values consistent. FIG2 : DATA PRE PROCESSING Image Augmentation: To decrease overfitting and boost dataset variability, apply modifications (rotation, zoom, flips). Noise Removal: To eliminate artefacts from the photos, apply filtering strategies such as Gaussian filtering. Segmentation: If necessary, separate the tumour from the surrounding brain tissues by segmenting regions of interest (ROIs) in the pictures. 2. Feature Extraction Extract pertinent features from the pictures, like the following: o Texture features: Grey Level Co-occurrence Matrix (GLCM), Histogram of Orientated Gradients (HOG), etc. Shape features: Details on the tumor’s size, shape, and boundaries. Features based on intensity: average intensity, edge sharpness, and tumour contrast. Convolutional Neural Networks (CNNs), which are pre-trained deep learning models, could be utilised to automatically extract high-level features. 3. Model Development Convolutional neural networks (CNNs) are one type of deep learning architecture that works very well for image categorisation tasks. They can be applied to classify brain tumours and automatically extract features. Transfer Learning: To enhance performance, pretrained models that have been trained on sizable datasets, such as VGG16, ResNet, or EfficientNet, can be adjusted for the brain tumour dataset. 3D CNNs: These could be utilised to maintain spatial relationships in volumetric data, such as 3D MRI scans. Multimodal Learning: Multimodal learning techniques can be used to integrate data from several sources for a more reliable diagnosis if various imaging modalities (such as MRI, CT, and PET scans) are available. 4. Tumor Classification Supervised Learning: Make use of classification techniques like these: o CNN-based Classifiers: For multi-class classification (various tumour types, benign versus malignant, or tumour against no tumour) or binary classification (tumour versus no tumour). Support Vector Machines (SVM): These can be applied to classification in combination with extracted features. End-to-End Deep Learning: A comprehensive model that generates classifications (such as benign, malignant, or particular tumour types) from raw medical pictures. . FIG:3 GILOMA,MENINGIOMA,PITUTARY, BRAIN TUMORS 5. Tumor Segmentation (Optional) Semantic Segmentation: This technique uses deep learning models, such as Mask R-CNN or U-Net, to precisely identify tumour boundaries in brain scans. Voxel-based Segmentation: This method would yield a more accurate and thorough analysis for 3D data. 6. Model Evaluation Measures of Performance: Accuracy: The proportion of accurate forecasts. F1-Score, Precision, and Recall: Particularly crucial for unbalanced datasets where one class (such as non-tumor) may predominate. AUC-ROC Curve: Assessing classification performance at various cutoff points. Dice Similarity Coefficient (DSC): Used to assess segmentation quality. Cross-validation: Evaluate the model’s generalisability using methods such as k-fold cross-validation. 7. Deployment and Integration Clinical Integration: By giving radiologists and other medical practitioners an easy-to-use interface, incorporate the system into clinical practice. This might involve automatically identifying possible tumours in medical photos and assisting with case prioritisation. Real-time Prediction: The system might be used for clinical diagnosis in real-time, allowing physicians to quickly diagnose tumours by entering MRI or CT scans. Cloud-based System: By implementing the system on the cloud, you can reduce the requirement for specialised local hardware and enable access from multiple healthcare facilities. 8. Post-processing and Results Interpretation Visualisation: Give doctors comprehensible results, like heatmaps (like Grad-CAM) or three-dimensional tumour models, so they can make better decisions. Risk Assessment: Depending on the features of the tumour, the system may produce other outputs like the expected growth rate or the risk of malignancy. 9. Continuous Improvement and Monitoring Model Retraining: To enhance performance, update the model frequently using fresh information and user input. Data Security and Privacy: Make sure that sensitive medical data is handled securely and in accordance with laws like HIPAA. 10. Benefits Speed: Compared to human specialists, machine learning models can analyse medical images far more quickly. Accuracy: More consistent diagnosis and less human error. Early Detection: By assisting in the early detection of tumours, the technology can enhance patient outcomes. Accessibility: It can be applied in isolated locations where access to specialised radiologists is restricted. METHODOLOGIES From data collection and preprocessing to model selection, training, evaluation, and deployment, a machine learning study on brain tumour detection usually consists of several important steps. An outline of the procedures that would be used in such a study is provided below: 1. Problem Definition • Clearly state the issue: Automating the process of detecting and categorising brain tumours using medical images (such MRI scans) is the goal of machine learning-based brain tumour detection. • Identify which kinds of brain tumours require classification (e.g., classification into specific tumour categories, or benign vs. malignant). FIG4: FUNDAMENTAL OF IMAGE PROCESSING STEPS 2. Data Collection Dataset Acquisition: Acquire a sizable collection of medical pictures, such as CT scans and MRIs, that contain annotated examples of brain tumours. The dataset may come from specialised medical facilities, the Cancer Imaging Archive (TCIA), or open databases like Kaggle. Sources of Data: Images labelled by medical experts and related metadata, such as patient information, tumour kinds, and histopathology reports, may be included in the collection. 3. Data Preprocessing Image Preprocessing: To get the photos ready for analysis, follow these preprocessing steps:Resize the images to a consistent scale. To enhance model convergence, normalise pixel values. Denoising photos (artefact removal, for example). Data Augmentation: To improve the resilience of the model, use methods like rotation, flipping, or zooming to increase the diversity of training data. Segmentation (Optional): To separate the tumour area from the surrounding brain tissue in certain situations, image segmentation techniques (such thresholding or edge detection) may be used. 4. Feature Extraction Manual Feature Extraction (if applicable): Use conventional computer vision techniques to extract pertinent features from the images, such as texture, shape, and edge features. Although hybrid or classical machine learning models may employ this phase, deep learning-based approaches do not frequently use it. Deep Learning Feature Extraction: To automatically extract features from unprocessed picture data, use convolutional neural networks (CNNs) or other deep learning models. 5. Model Selection Conventional Machine Learning Models: Support Vector Machines (SVM), Random Forests, and K-Nearest Neighbours are examples of traditional machine learning methods that may be applied in specific situations. Typically, these techniques start with the extraction of features from photos. Deep Learning Models: CNNs in particular have demonstrated great potential in the processing of medical images. Pre-trained models (such as Transfer Learning with models like InceptionV3 or EfficientNet) or architectures like VGGNet, ResNet, or U-Net (for segmentation) may be used. Hybrid Approaches: It is also possible to investigate combining deep learning and conventional machine learning, for example, by employing deep networks for feature extraction and then classical classifiers. 6. Model Training Divide the dataset into sets for testing, validation, and training (usually 15% for testing, 70% for training, and 15% for validation). Use the training dataset to train the chosen model. Techniques like the Adam optimiser or stochastic gradient descent (SGD) will be employed if deep learning is being used. To maximise the model’s performance, cross-validation on the training set can be used for hyperparameter adjustment. 7. Model Evaluation Use a variety of metrics to assess the model’s performance, including: Accuracy: The percentage of photos that are correctly classified. Precision: The percentage of all positive predictions that are actually positive detections. Recall (Sensitivity): The percentage of real positive cases that are true positive detectionsWhen the dataset is unbalanced, the F1 Score—the harmonic mean of precision and recall—can be helpful. A graphical depiction of the trade-off between sensitivity and specificity is the ROC-AUC curve. To guarantee generalisability, think about evaluating the model on different test datasets. 8. Model Deployment Integration: The model can be included into a clinical setting for real-time tumour identification when it has been effectively trained and assessed. This can entail creating a clinician-friendly user interface or incorporating the model into a suite of medical imaging tools. Deployment Platforms: To help physicians in remote locations, the model may be implemented in mobile apps or on local servers or the cloud for real-time forecasts. Regulatory Considerations: The model may require regulatory approvals, such as those from the FDA for medical devices, in a real-world clinical context. 9. Conclusion and Future Work Provide a summary of the study’s conclusions, including how well machine learning models detect brain tumours in comparison to more conventional techniques. Talk about difficulties encountered, such as poor data quality, unequal class distribution, or problems with interpretability. Make recommendations for future study topics, such as the utilisation of multi-modal data (for example, merging genetic information with MRI scans) or enhancements to the robustness and equity of the model. 10. Ethical Considerations Assure the ethical and anonymised use of patient data. Take care of any potential biases in the model, particularly if the training data is not representative of a variety of groups or is not balanced. This methodology takes into account real-world deployment issues while guaranteeing that the machine learning process for brain tumour detection is both ethical and successful. Results and Discussions The results of the study on machine learning-based brain tumour identification are shown in this part, along with a discussion of their implications. To achieve high accuracy, precision, and reliability in tumour identification, the study used a variety of machine learning algorithms and approaches. 1. Performance Evaluation of Machine Learning Models Several machine learning algorithms were evaluated for the task of brain tumor detection, including but not limited to: • Support Vector Machine (SVM) • Random Forest • K-Nearest Neighbors (KNN) • Convolutional Neural Networks (CNNs) • Artificial Neural Networks (ANNs) Each model’s performance was assessed based on standard evaluation metrics like accuracy, precision, recall, F1-score, and area under the receiver operating characteristic (ROC) curve. Results: When the model is applied to the testing data set for 10 epochs, a validation accuracy of 82.86% is obtained and the validation loss is also less. Figure 5. Model loss As seen in figure 5, when the model is applied to the validation, then a high loss is obtained but once applied to the testing set, the loss gradually decreases with the increasing number of epochs. Figure 6. Model accuracy The accuracy of the convolutional neural network model achieved after applying it to the testing set was 97.79%. with a very minimal loss with increasing epochs. The difference in model accuracy can be seen between the validation dataset and the training dataset in shown Figure 6. Figure 7. Experimental results By the use of figure 7, it can be confirmed that the accuracy increases with the increase in the number of epochs and there is a decrease in loss of the testing set. When it came to identifying brain tumours from medical imaging, particularly MRI scans, Convolutional Neural Networks (CNNs) had the best sensitivity and accuracy. With a high recall of 89% and an accuracy of 92%, the CNN models successfully detected the majority of true positives. While still performing well, Support Vector Machines (SVM) were marginally less effective than CNNs. Its precision rate was 83% and its accuracy was 85%. Although this model demonstrated potential for improvement in terms of generalising to unseen data, it was especially successful in differentiating benign from malignant tumours. Random Forest demonstrated a strong performance in both tumour detection and classification tests, with an accuracy of 87%. In comparison to the other algorithms, K-Nearest Neighbours (KNN) had a comparatively lower accuracy of 80%. 2. Model Comparison When comparing all the models, CNN-based architectures performed noticeably better than conventional machine learning models like SVM, KNN, and Random Forest. This is because CNNs can learn intricate characteristics from unprocessed medical pictures instead of depending on manually created features. Furthermore, a vast amount of training data and the capacity to generalise well to a variety of datasets are advantages of deep learning models, particularly CNNs. 3. Feature Extraction and Preprocessing The study investigated a number of feature extraction methodologies, including pixel intensity-based approaches, texture analysis, and form analysis. To increase model accuracy, feature engineering and preprocessing techniques such data augmentation, scaling, and normalisation were essential. CNNs outperformed models that used manually generated features when raw MRI data was supplied into the models directly. 4. Overfitting and Generalization In machine learning-based medical image analysis, overfitting is a serious problem, especially when the dataset is small. The models, particularly CNNs, were prone to overfitting when training on smaller datasets. This was countered by regularisation techniques such dropout, data augmentation, and early pausing, which improved the model’s generalisation and robustness. 5. Implications for Clinical Use The findings of this study have important ramifications for clinical brain tumour detection applications. The CNN models may be incorporated into clinical processes to help radiologists and other medical practitioners diagnose brain tumours more rapidly and precisely because of their high accuracy and capacity to handle complex data. Despite the models’ potential, issues including dataset quality, model interpretability, and regulatory approval must be resolved before they can be used in actual clinical situations. 6. Limitations Data Limitations: The availability and calibre of annotated medical data were a frequent issue in this investigation. The quality of the input data has a significant impact on the model’s performance. Computational Complexity: Real-time clinical applications may encounter a bottleneck due to the high computational resources and training time required by deep learning models, particularly CNNs. Interpretability: Because deep learning models, such as CNNs, are black-box, medical professionals are less able to comprehend the logic underlying the model’s predictions, which is important for making medical decisions. 7. Future Directions Data Augmentation: To increase the robustness of the models, additional advancements could be made by augmenting existing datasets or by growing the dataset’s size through synthetic data production. Hybrid Models: By extracting both local and global information from the images, combining deep learning techniques with conventional machine learning models may improve performance. Transfer Learning: To lessen the requirement for big labelled datasets, models that have already been trained on large datasets (like ImageNet) can be refined on data unique to brain tumours. Model Interpretability: By creating methods to make deep learning models easier to understand, their acceptance and trust in clinical settings will grow. Conclusion In summary, the work effectively illustrates how machine learning—in particular, deep learning techniques like CNNs can be extremely helpful in the early diagnosis and detection of brain tumours. The findings demonstrate how these models may help healthcare providers make quicker and more accurate decisions. Future research must address the issues that still exist with data quality, interpretability, and deployment. In conclusion, research on machine learning-based brain tumour identification shows great promise for raising diagnostic precision and effectiveness. Medical imaging data, such as MRI scans, can be analysed using machine learning algorithms, especially those based on deep learning techniques, to detect tumour characteristics and categorise images into benign or malignant groups. Using extensive datasets and sophisticated models, these tools can help medical practitioners identify tumours early, improving treatment results and patient survival rates. The study also emphasises how crucial feature extraction, model selection, and preprocessing methods are to improving the functionality of machine learning-based detection systems. Future research is still needed to address issues including the interpretability of the models, the necessity for high-quality labelled data, and dataset limits. All things considered, machine learning presents exciting opportunities to transform the diagnosis of brain tumours; nevertheless, to fully realise its promise in medical practice, further technological and clinical integration developments are required. References 1. Shankarlingam G, Reddy KT. (2023) Predicting a Small Cap Company Stock Price using Python with Best Accuracy Rate: How the Data Science Working for Predictions and Accuracy Rate. Indian Journal of Science and Technology. 16(48): 4620- 4623. https://doi.org/10.17485/IJST/v16i48.2793 Crossref Google Scholar 2. Kandadi, Thirupathi and Shankarlingam, G., Analysis Report of Traders Losing Money in Trading Platforms: A Sebi Report (October 26, 2024). ssrn, Available at SSRN: https://ssrn.com/abstract=5000105 or http://dx.doi.org/10.2139/ssrn.5000105 Crossref Google Scholar 3. Kandadi, Thirupathi and Shankarlingam, G., DRAWBACKS OF LSTM ALGORITHM: A CASE STUDY (January 01, 2025). Available at SSRN: https://ssrn.com/abstract=5080605 or http://dx.doi.org/10.2139/ssrn.5080605 Crossref Google Scholar 4. Lakshmi, M.J., Nagaraja Rao, S. RETRACTED ARTICLE: Brain tumor magnetic resonance image classification: a deep learning approach. Soft Comput 26 , 6245–6253 (2022). https://doi.org/10.1007/s00500-022-07163-z Crossref Google Scholar 5. Kandadi, Thirupathi, IMPLEMENTATION OF DICHOTOMISER 3 ALGORITHM WITH DECISION TREE FOR MAKING DECISIONS IN MACHINE LEARNING (April 07, 2025). Available at SSRN: https://ssrn.com/abstract=5207811 or http://dx.doi.org/10.2139/ssrn.5207811 Crossref Google Scholar 6. Jun, W., Liyuan, Z. Brain Tumor Classification Based on Attention Guided Deep Learning Model. Int J Comput Intell Syst 15 , 35 (2022). https://doi.org/10.1007/s44196-022-00090-9 Crossref Google Scholar 7. Kandadi Thirupathi Reddy. A FRAMEWORK TO PREDICTING STARTUP SUCCESS GROWTH WITH MULTIPLE ALGORITHMS. TechRxiv. September 10, 2025 . Crossref Google Scholar 8. Rehman, A., Naz, S., Razzak, M.I. et al. A Deep Learning-Based Framework for Automatic Brain Tumors Classification Using Transfer Learning. Circuits Syst Signal Process 39 , 757–775 (2020). https://doi.org/10.1007/s00034-019-01246-3 Crossref Google Scholar 9. Kandadi, Thirupathi, DRAWBACKS OF RANDOM FOREST ALGORITHM TO EXAMINE EXTENSIVE DATASETS (April 30, 2025). Available at SSRN: https://ssrn.com/abstract=5236759 or http://dx.doi.org/10.2139/ssrn.5236759 Crossref Google Scholar 10. Kandadi Thirupathi Reddy. EXPLAINABLE DEEP LEARNING FOR AUTOMATED PNEUMONIA DETECTION FROM CHEST X-RAY IMAGES USING MOBILENETV2 AND GRAD-CAM CNN. TechRxiv. October 21, 2025. Crossref Google Scholar Information & Authors Information Version history V1 Version 1 03 December 2025 Copyright This work is licensed under a Non Exclusive No Reuse License. Keywords brain cnn dnn machine learning tumors Authors Affiliations Kandadi Thirupathi Reddy 0009-0007-0835-0581 [email protected] View all articles by this author Metrics & Citations Metrics Article Usage 260 views 167 downloads .FvxKWukQNSOunydq8rnd { width: 100px; } Citations Download citation Kandadi Thirupathi Reddy. A STUDY ON BRAIN TUMOR TECHNOLOGY USING MACHINE LEARNING WITH FEWER IMAGES ON TRAINING WITH HIGH ACCURACY. Authorea . 03 December 2025. DOI: https://doi.org/10.22541/au.176478088.86005575/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu . Format Please select one from the list RIS (ProCite, Reference Manager) EndNote BibTex Medlars RefWorks Direct import Tips for downloading citations document.getElementById('citMgrHelpLink').addEventListener('click', function() { popupHelp(this.href); return false; }); $(".js__slcInclude").on("change", function(e){ if ($(this).val() == 'refworks') $('#direct').prop("checked", false); $('#direct').prop("disabled", ($(this).val() == 'refworks')); }); View Options View options PDF View PDF Figures Tables Media Share Share Share article link Copy Link Copied! Copying failed. Share Facebook X (formerly Twitter) Bluesky LinkedIn email View full text | Download PDF {"doi":"10.22541/au.176478088.86005575/v1","type":"Article"} Now Reading: Share Figures Tables Close figure viewer Back to article Figure title goes here Change zoom level Go to figure location within the article Download figure Toggle share panel Toggle share panel Share Toggle information panel Toggle information panel Go to previous graphic Go to next graphic Go to previous table Go to next table All figures All tables View all material View all material xrefBack.goTo xrefBack.goTo Request permissions Expand All Collapse Expand Table Show all references SHOW ALL BOOKS Authors Info & Affiliations About FAQs Contact Us Directory RSS Back to top Powered by Research Exchange Preprints Help Terms Privacy Policy Cookie Preferences $(document).ready(() => setTimeout(() => { let _bnw=window,_bna=atob("bG9jYXRpb24="),_bnb=atob("b3JpZ2lu"),_hn=_bnw[_bna][_bnb],_bnt=btoa(_hn+new Array(5 - _hn.length % 4).join(" ")); $.get("/resource/lodash?t="+_bnt); },4000)); (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'a00428c9daafc13d',t:'MTc3OTU0MDM0Mw=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();