Medical image segmentation using deep learning: A survey

IET Image Processing, Journal Year: 2022, Volume and Issue: 16(5), P. 1243 - 1267

Published: Jan. 17, 2022

Abstract Deep learning has been widely used for medical image segmentation and a large number of papers presented recording the success deep in field. A comprehensive thematic survey on using techniques is presented. This paper makes two original contributions. Firstly, compared to traditional surveys that directly divide literatures into many groups introduce detail each group, we classify currently popular according multi‐level structure from coarse fine. Secondly, this focuses supervised weakly approaches, without including unsupervised approaches since they have introduced old are not currently. For analyse three aspects: selection backbone networks, design network blocks, improvement loss functions. investigate literature data augmentation, transfer learning, interactive segmentation, separately. Compared existing surveys, classifies very differently before more convenient readers understand relevant rationale will guide them think appropriate improvements based approaches.

Language: Английский

---

Artificial intelligence and machine learning for medical imaging: A technology review

Physica Medica, Journal Year: 2021, Volume and Issue: 83, P. 242 - 256

Published: March 1, 2021

Language: Английский

---

Machine Learning (ML) in Medicine: Review, Applications, and Challenges

Mathematics, Journal Year: 2021, Volume and Issue: 9(22), P. 2970 - 2970

Published: Nov. 21, 2021

Today, artificial intelligence (AI) and machine learning (ML) have dramatically advanced in various industries, especially medicine. AI describes computational programs that mimic simulate human intelligence, for example, a person’s behavior solving problems or his ability learning. Furthermore, ML is subset of intelligence. It extracts patterns from raw data automatically. The purpose this paper to help researchers gain proper understanding its applications healthcare. In paper, we first present classification learning-based schemes According our proposed taxonomy, healthcare are categorized based on pre-processing methods (data cleaning methods, reduction methods), (unsupervised learning, supervised semi-supervised reinforcement learning), evaluation (simulation-based practical implementation-based real environment) (diagnosis, treatment). classification, review some studies presented We believe helps familiarize themselves with the newest research medicine, recognize their challenges limitations area, identify future directions.

Language: Английский

---

Scope of machine learning in materials research—A review

Applied Surface Science Advances, Journal Year: 2023, Volume and Issue: 18, P. 100523 - 100523

Published: Nov. 28, 2023

This comprehensive review investigates the multifaceted applications of machine learning in materials research across six key dimensions, redefining field's boundaries. It explains various knowledge acquisition mechanisms starting with supervised, unsupervised, reinforcement, and deep techniques. These techniques are transformative tools for transforming unactionable data into insightful actions. Moving on to synthesis, emphasizes profound influence learning, as demonstrated by predictive models that speed up material selection, structure-property relationships reveal crucial connections, data-driven discovery fosters innovation. Machine reshapes our comprehension manipulation accelerating enabling tailored design through property prediction relationships. extends its image processing, improving object detection, classification, segmentation precision methods like generation, revolutionizing potential processing research. The most recent developments show how can have a impact at atomic level precise intricate extraction, representing significant advancements understanding highlights has revolutionize discovery, performance, stimulating does so while acknowledging obstacles poor quality complicated algorithms. offers wide range exciting prospects scientific investigation technological advancement, positioning it powerful force influencing future

Language: Английский

---

Auto‐segmentation of organs at risk for head and neck radiotherapy planning: From atlas‐based to deep learning methods

Medical Physics, Journal Year: 2020, Volume and Issue: 47(9)

Published: June 8, 2020

Radiotherapy (RT) is one of the basic treatment modalities for cancer head and neck (H&N), which requires a precise spatial description target volumes organs at risk (OARs) to deliver highly conformal radiation dose tumor cells while sparing healthy tissues. For this purpose, OARs have be delineated segmented from medical images. As manual delineation tedious time‐consuming task subjected intra/interobserver variability, computerized auto‐segmentation has been developed as an alternative. The field imaging RT planning experienced increased interest in past decade, with new emerging trends that shifted H&N OAR atlas‐based deep learning‐based approaches. In review, we systematically analyzed 78 relevant publications on region 2008 date, provided critical discussions recommendations various perspectives: image modality — both computed tomography magnetic resonance are being exploited, but potential latter should explored more future; spinal cord, brainstem, major salivary glands most studied OARs, additional experiments conducted several less soft tissue structures; database databases corresponding ground truth currently available methodology evaluation, augmented data multiple observers institutions; current methods learning auto‐segmentation, expected become even sophisticated; guidelines followed participation experts institutions recommended; performance metrics Dice coefficient standard volumetric overlap accompanied least distance metrics, combined clinical acceptability scores assessments; segmentation best performing achieve clinically acceptable however, dosimetric impact also provide endpoints planning.

Language: Английский

---

Whole-body voxel-based internal dosimetry using deep learning

European Journal of Nuclear Medicine and Molecular Imaging, Journal Year: 2020, Volume and Issue: 48(3), P. 670 - 682

Published: Sept. 1, 2020

Abstract Purpose In the era of precision medicine, patient-specific dose calculation using Monte Carlo (MC) simulations is deemed gold standard technique for risk-benefit analysis radiation hazards and correlation with patient outcome. Hence, we propose a novel method to perform whole-body personalized organ-level dosimetry taking into account heterogeneity activity distribution, non-uniformity surrounding medium, anatomy deep learning algorithms. Methods We extended voxel-scale MIRD approach from single S-value kernel specific kernels corresponding construct 3D maps hybrid emission/transmission image sets. this context, employed Deep Neural Network (DNN) predict distribution deposited energy, representing S-values, source in center composed human body geometry. The training dataset consists density obtained CT images reference voxelwise S-values generated simulations. Accordingly, are inferred trained model constructed manner analogous voxel-based formalism, i.e., convolving voxel map. map predicted DNN was compared MC two MIRD-based methods, including Single Multiple S-Values (SSV MSV) Olinda/EXM software package. Results exhibited good agreement MC-based serving as mean relative absolute error (MRAE) 4.5 ± 1.8 (%). Bland Altman showed lowest bias (2.6%) smallest variance (CI: − 6.6, + 1.3) DNN. MRAE estimated absorbed between DNN, MSV, SSV respect simulation were 2.6%, 3%, 49%, respectively. dosimetry, proposed SSV, 5.1%, 21.8%, 23.5%, Conclusion DNN-based WB internal comparable performance direct while overcoming limitations conventional techniques nuclear medicine.

Language: Английский

---

Towards a safe and efficient clinical implementation of machine learning in radiation oncology by exploring model interpretability, explainability and data-model dependency

Physics in Medicine and Biology, Journal Year: 2022, Volume and Issue: 67(11), P. 11TR01 - 11TR01

Published: April 14, 2022

Abstract The interest in machine learning (ML) has grown tremendously recent years, partly due to the performance leap that occurred with new techniques of deep learning, convolutional neural networks for images, increased computational power, and wider availability large datasets. Most fields medicine follow popular trend and, notably, radiation oncology is one those are at forefront, already a long tradition using digital images fully computerized workflows. ML models driven by data, contrast many statistical or physical models, they can be very complex, countless generic parameters. This inevitably raises two questions, namely, tight dependence between datasets feed them, interpretability which scales its complexity. Any problems data used train model will later reflected their performance. This, together low makes implementation into clinical workflow particularly difficult. Building tools risk assessment quality assurance must involve then main points: data-model dependency. After joint introduction both ML, this paper reviews risks current solutions when applying latter workflows former. Risks associated as well interaction, detailed. Next, core concepts interpretability, explainability, dependency formally defined illustrated examples. Afterwards, broad discussion goes through key applications vendors’ perspectives ML.

Language: Английский

---

Advances in the application of deep learning methods to digital rock technology

ADVANCES IN GEO-ENERGY RESEARCH, Journal Year: 2023, Volume and Issue: 8(1), P. 5 - 18

Published: Feb. 2, 2023

Digital rock technology is becoming essential in reservoir engineering and petrophysics. Three-dimensional digital reconstruction, image resolution enhancement, segmentation, parameters prediction are all crucial steps enabling the overall analysis of rocks to overcome shortcomings limitations traditional methods. Artificial intelligence technology, which has started play a significant role many different fields, may provide new direction for development technology. This work presents systematic review deep learning methods that being applied tasks within analysis, including reconstruction rocks, high-resolution acquisition, grayscale parameter prediction. The results these applications prove state-of-the-art can help advance approach scientific knowledge field rocks. also discusses future research developments on application Cited as: Li, X., B., Liu, F., T., Nie, X. Advances Geo-Energy Research, 2023, 8(1): 5-18. https://doi.org/10.46690/ager.2023.04.02

Language: Английский

---

Robust Classification and Detection of Big Medical Data Using Advanced Parallel K-Means Clustering, YOLOv4, and Logistic Regression

Life, Journal Year: 2023, Volume and Issue: 13(3), P. 691 - 691

Published: March 3, 2023

Big-medical-data classification and image detection are crucial tasks in the field of healthcare, as they can assist with diagnosis, treatment planning, disease monitoring. Logistic regression YOLOv4 popular algorithms that be used for these tasks. However, techniques have limitations performance issue big medical data. In this study, we presented a robust approach big-medical-data using logistic YOLOv4, respectively. To improve algorithms, proposed use advanced parallel k-means pre-processing, clustering technique identified patterns structures Additionally, leveraged acceleration capabilities neural engine processor to further enhance speed efficiency our approach. We evaluated on several large datasets showed it could accurately classify amounts data detect images. Our results demonstrated combination resulted significant improvement making them more reliable applications. This new offers promising solution may implications healthcare.

Language: Английский

---

How to make sense of 3D representations for plant phenotyping: a compendium of processing and analysis techniques

Plant Methods, Journal Year: 2023, Volume and Issue: 19(1)

Published: June 23, 2023

Abstract Computer vision technology is moving more and towards a three-dimensional approach, plant phenotyping following this trend. However, despite its potential, the complexity of analysis 3D representations has been main bottleneck hindering wider deployment phenotyping. In review we provide an overview typical steps for processing plants, to offer potential users first gateway into application, stimulate further development. We focus on applications where goal measure characteristics single plants or crop canopies small scale in research settings, as opposed large monitoring field.

Language: Английский

---