RESUMO
Single-molecule detection (SMD) holds considerable promise in biomedical research. Although atomic force microscopy (AFM) provides an important technique with nanoscale resolution for SMD, its broader application is limited by labeling challenges and slow data processing. Here, we present a machine learning (ML)-powered strategy combining AFM and DNA nanotags for SMD and cancer diagnosis. Nickases are applied to create specific single-strand breaks in target DNA, allowing insertion of exogenous DNA to attach shape-distinct nanotags for AFM imaging. A YOLOv5l algorithm is adopted to automatically recognize target objects in AFM images, which can classify 370 structures in 1.21 seconds with 98% accuracy. The proof of concept of this strategy is confirmed by identifying nickase-edited sites on both linear and circular DNA. Its practical applicability is demonstrated by detecting KRAS Gly12Arg (G12R) and p53 Arg175His (R175H) mutations in samples from patients with pancreatic and colorectal cancer, with accuracy rivaling Sanger sequencing and quantitative polymerase chain reaction, opening avenues for SMD.
Assuntos
DNA , Aprendizado de Máquina , Neoplasias , Neoplasias Pancreáticas , Imageamento de Moléculas Individualizadas , Humanos , Microscopia de Força Atômica/métodos , DNA/química , Imageamento de Moléculas Individualizadas/métodos , Neoplasias/diagnóstico , Neoplasias/genética , Proteína Supressora de Tumor p53/genética , Mutação , Proteínas Proto-Oncogênicas p21(ras)/genética , Algoritmos , Neoplasias Pancreáticas/diagnóstico , Neoplasias Pancreáticas/genética , Neoplasias Colorretais/diagnóstico , Neoplasias Colorretais/genéticaRESUMO
Optical coherence tomography (OCT) is a widely used imaging modality for diagnosing and monitoring macular diseases, including diabetic macular edema (DME) and choroidal neovascularization (CNV), both of which can cause severe visual impairment. Clinicians rely on various OCT biomarkers to identify these conditions. An algorithm was developed in Python to extract biomarker-associated features from OCT images and applied to a pre-labeled dataset containing normal, DME, and CNV images. Distribution analysis confirmed that the extracted features aligned with the existing literature. Using these features, LightGBM classified the OCT images, achieving 91% accuracy and 98% area under the receiver operating characteristic curve. Based on these promising results, this algorithm could contribute to the development of more advanced feature extraction methodologies for the diagnosis of macular diseases using traditional machine learning approaches. Such algorithms could potentially be integrated into automated patient screening systems.
Assuntos
Neovascularização de Coroide , Retinopatia Diabética , Edema Macular , Tomografia de Coerência Óptica , Humanos , Tomografia de Coerência Óptica/métodos , Edema Macular/diagnóstico por imagem , Edema Macular/classificação , Retinopatia Diabética/diagnóstico por imagem , Retinopatia Diabética/classificação , Neovascularização de Coroide/diagnóstico por imagem , Neovascularização de Coroide/classificação , Algoritmos , Aprendizado de Máquina , Curva ROC , Interpretação de Imagem Assistida por Computador/métodosRESUMO
Genome-wide association studies (GWAS) have revolutionized our understanding of genetic contributions to complex diseases by identifying single-nucleotide polymorphisms (SNPs) associated with disease predisposition. Despite the substantial progress made in identifying risk factors for conditions like cancer and cardiovascular diseases, interpreting the functional impact of identified variants remains a challenge, particularly when silent mutations are involved. Silent mutations, once considered irrelevant to disease mechanisms, have emerged as significant players influencing mRNA formation, splicing, and translation processes. This study utilized the Genetic Association Database (GAD) to analyze and identify the significance of silent mutations across a wide range of diseases, employing advanced machine learning techniques and the Apriori algorithm to extract association rules from a biomedical dataset. The Apriori algorithm was applied to identify strong correlations between diseases and chromosomes, using parameters such as support, confidence, and lift to evaluate the strength and importance of these associations. Our results demonstrated the capability of the Apriori algorithm to uncover biologically meaningful relationships, which could be instrumental in improving our understanding of genetic predispositions and guiding precision medicine efforts. These findings underscore the importance of silent mutations in disease etiology and highlight the potential of bioinformatics tools in unraveling complex genetic interactions.
Assuntos
Algoritmos , Bases de Dados Genéticas , Estudo de Associação Genômica Ampla , Polimorfismo de Nucleotídeo Único , Mutação Silenciosa , Humanos , Estudo de Associação Genômica Ampla/métodos , Predisposição Genética para Doença , Aprendizado de Máquina , Biologia Computacional/métodosRESUMO
Leukemia is a blood cancer characterized by the abnormal proliferation of blood cells and chronic myeloid leukemia (CML) is a type of leukemia disease that is noted by inappropriate increase of leukemic stem cell in hematopoietic niche. Mathematical modeling gives a strict conceptualization of disease development and the assessment of possible therapeutic reactions. This paper introduces the comparative study and analysis of the three known models of CML including Niche competition, Niche independent, and partial Niche dependence (PND). In case of every model, we determine the equilibrium points, calculate the basic reproduction number R 0 , and examine the local stability. We can compare the mathematical frameworks of the models, even though they have mathematical differences, and identify the essence of the role that niche interactions play in the persistence of leukemia. Numerical simulations, which were conducted by the fourth-order Runge-Kutta (RK-4) method, demonstrate the threshold dynamics of and shows clear qualitative behavior in different models. PND model proves to be the most biologically consistent behavior which is consistent with the known CML niche ecology. Our results give us a coherent understanding of CML modeling and a solid basis of the PND framework on the development of more clinically sound models that consider immune interactions, effects of treatment, and patient heterogeneity.
Assuntos
Leucemia Mielogênica Crônica BCR-ABL Positiva , Modelos Biológicos , Leucemia Mielogênica Crônica BCR-ABL Positiva/patologia , Humanos , Simulação por Computador , Células-Tronco Neoplásicas/patologia , Número Básico de Reprodução , Nicho de Células-Tronco , Células-Tronco Hematopoéticas/patologia , Proliferação de Células , AlgoritmosRESUMO
Ophthalmology is undergoing rapid transformation through the integration of smart technologies such as artificial intelligence (AI), big data analytics, and clinical decision support systems (CDSS). With increasing pressure to improve clinical efficiency and manage growing patient volumes, the potential for smart technologies to streamline ophthalmic care warrants more exploration. To date, smart technologies have demonstrated potential as practical adjunctive tools that support ophthalmic referrals and clinical practice in ophthalmology. Smart technologies that support ophthalmic referrals now include CDSS that contain algorithms with the capacity to more efficiently identify suspected ophthalmic diseases that may be urgent or require prompt treatment in the primary care setting, compared with traditional referral models. These approaches also include installation of AI-powered ophthalmic imaging machines and electronic health records-analytical packages in primary care offices, where they can be used to screen for structural, historical, or symptomatic manifestations of ophthalmic diseases requiring ophthalmologist evaluation. Meanwhile, smart technologies that support ophthalmology practices include AI and big data simulations for optimized patient encounter schedules and chatbot-facilitated appointment confirmations. Amidst a smart technology renaissance, review is needed to capture existing smart technologies to inform integration in the practices of ophthalmic and general practitioners. This article aims to review the clinical utility of emerging smart technology relevant to ophthalmic referrals and ophthalmology practice.
Assuntos
Inteligência Artificial , Big Data , Oftalmologia , Encaminhamento e Consulta , Humanos , Inteligência Artificial/tendências , Oftalmologia/métodos , Oftalmologia/tendências , Encaminhamento e Consulta/tendências , Sistemas de Apoio a Decisões Clínicas , Oftalmopatias/diagnóstico , Registros Eletrônicos de Saúde , AlgoritmosRESUMO
BACKGROUND: MR-guided adaptive radiotherapy (ART) allows for daily plan optimization based on patient-specific anatomy. Accumulated doses, driven by deformable image registration (DIR), of daily fractions can provide cumulative dose metrics and insights into toxicity and tumor control. In prostate ART, inter- and intra-factional deformations, particularly due to bladder and rectum, pose a challenge to accurate DIR generation. PURPOSE: To quantify geometric and dosimetric accuracy of a proposed prostate MR-to-MR DIR approach to support MR-guided ART dose accumulation. METHODS: We evaluated DIR accuracy in 25 patients treated with 30 Gy in five fractions on a 1.5 T MR-linac using an adaptive workflow. For all patients, a reference MR was used for planning, with three images collected at each fraction: adapt MR for adaptive planning, verify MR for pretreatment position verification and beam-on for capturing anatomy during radiation delivery. We assessed three DIR approaches: intensity-based, intensity-based with controlling structures (CS), and intensity-based with controlling structures and points of interest (CS + P). DIRs were performed between the reference and fraction images and within fractions (adapt-to-verify and adapt-to-beam-on). For the evaluation, we propagated CTV, bladder, and rectum contours using the DIRs and compared each to manually delineated contours using Dice similarity coefficient, mean distance to agreement, and dose-volume metrics. RESULTS: CS and CS + P improved geometric agreement between manual and propagated contours over intensity-only DIR. For example, mean distance to agreement (DTAmean) for reference-to-beam-on intensity-only DIR was 0.131 ± 0.009 cm (CTV), 0.46 ± 0.08 cm (bladder), and 0.154 ± 0.013 cm (rectum). For the CS, the DTAmean values were 0.018 ± 0.002, 0.388 ± 0.14, and 0.036 ± 0.013 cm. Finally, for CS + P, these values were 0.015 ± 0.001, 0.025 ± 0.004, and 0.021 ± 0.002 cm. Dosimetrically, comparing CS and CS + P for reference to beam-on DIRs resulted in a change of CTV D98% from [-29 cGy, 19 cGy] to [-18 cGy, 26 cGy], bladder D5cc from [-51 cGy, 544 cGy] to [-79 cGy, 36 cGy], and rectum D1cc from [-106 cGy, 72 cGy] to [-52 cGy, 74 cGy]. CONCLUSION: CS improved geometric and dosimetric accuracy over intensity-only DIR, with CS + P providing further performance improvement, particularly for bladder. However, session image segmentation remains a challenge, which may be addressed with automated contouring.
Assuntos
Processamento de Imagem Assistido por Computador , Imageamento por Ressonância Magnética , Neoplasias da Próstata , Planejamento da Radioterapia Assistida por Computador , Radioterapia Guiada por Imagem , Humanos , Masculino , Neoplasias da Próstata/radioterapia , Neoplasias da Próstata/diagnóstico por imagem , Planejamento da Radioterapia Assistida por Computador/métodos , Radioterapia Guiada por Imagem/métodos , Dosagem Radioterapêutica , Imageamento por Ressonância Magnética/métodos , Radioterapia de Intensidade Modulada/métodos , Órgãos em Risco/efeitos da radiação , Processamento de Imagem Assistido por Computador/métodos , AlgoritmosRESUMO
BACKGROUND: Quantum computing (QC) is emerging as a transformative tool for solving complex optimization problems across various fields, including biomedical applications. While classical optimization methods are well-established, they frequently face limitations when applied to complex and large-scale problems in radiotherapy planning. PURPOSE: This study aims to explore the implementation and evaluate the effectiveness of quantum optimization methods, specifically quantum annealing and the quantum approximate optimization algorithm (QAOA), in radiotherapy planning. In particular, we employ an Ising Hamiltonian formulation of the cost function, empirically implementing it on annealing-based quantum hardware and for the first time on circuit-based one. METHODS: We formulated a simplified radiotherapy optimization problem and solved it using quantum annealing on a D-Wave quantum annealer. Subsequently, we adapted this optimization problem for the QAOA framework and implemented it on IBM Quantum circuit-model hardware. Comparative analyses were conducted between classical and quantum methods and implementations, highlighting QC's potential advantages and limitations in specific optimization contexts. To demonstrate that the Hamiltonian formulation is valid and practically usable, we first tested it in simplified proof-of-principle examples and then extended it to a more clinically relevant bilateral prostate proton plan. In this new example, dose parameters were extracted directly from a commercial treatment planning system (RayStation) and incorporated into the Hamiltonian optimization workflow. Both one-qubit-per-voxel and two-qubits-per-voxel encodings were evaluated to illustrate scalability. Additionally, we discussed scalability considerations, practical challenges, and future research directions necessary for integrating quantum algorithms into routine clinical radiation therapy practices. RESULTS: To our knowledge, this study presents the first demonstration of using QC circuit-model hardware for radiotherapy planning optimization. The quantum annealing approach successfully determined the optimal solution. Convergence was achieved after 20 iterations on a quantum simulator (noise free) and after 100 iterations on actual quantum hardware (due to inherent hardware noise). In the bilateral prostate proton plan derived from realistic data, the Hamiltonian-based optimization assigned higher dose to the prostate relative to surrounding organs-at-risk, confirming the feasibility of applying QC optimization directly to clinically sourced parameters. CONCLUSIONS: Quantum optimization techniques demonstrate potential advantages over classical methods, particularly in complex optimization scenarios relevant to radiation therapy. The formulation of a Hamiltonian cost function, its validation on real quantum hardware, and its application to realistic data collectively represent a first concrete step toward QC-based treatment planning optimization in medical physics. Future research should focus on addressing scalability, overcoming practical implementation challenges, and advancing the development of scalable, fault-tolerant quantum systems suitable for clinical integration.
Assuntos
Teoria Quântica , Planejamento da Radioterapia Assistida por Computador , Planejamento da Radioterapia Assistida por Computador/métodos , Humanos , Algoritmos , Dosagem Radioterapêutica , Neoplasias da Próstata/radioterapiaRESUMO
Medical imaging approaches widely employ deep neural networks for the investigation and diagnosis of different skin disorders. However, recent studies suggest that even a proficient model based on deep learning might struggle with generalization when applied to datasets from disparate cohorts due to domain shift phenomena. Meanwhile, there are usually need many well-labelled images utilized for the training process to attain a stronger level of performance. In order to alleviate the domain shift and the necessity for adequate training data, we introduce a novel method termed as adversarial selective domain adaption with feature cluster (ASDA). It achieves effective performance improvement of model when the target dataset is smaller than the source dataset. Specifically, we generate a set of feature clusters for each sample in the target domain to alleviate the demand for data. Subsequently, a conditional domain adversarial network is used to mitigate domain shift. Finally, due to consistency issues between feature clusters and samples, we propose a method of selective minmax entropy to maintain consistency. Our method diverges from typical domain adaption approaches that solely target reducing the domain gap. Instead, we address both the discrepancy between domains and the problem of limited data in the target dataset simultaneously. Extensive experiments have been undertaken on datasets pertaining to skin cancer, that confirms ASDA's efficacy in skin cancer diagnosis for dermatoscopic and clinic image.
Assuntos
Neoplasias Cutâneas , Humanos , Neoplasias Cutâneas/diagnóstico , Neoplasias Cutâneas/diagnóstico por imagem , Aprendizagem Profunda , Redes Neurais de Computação , Algoritmos , Análise por ConglomeradosRESUMO
BACKGROUND: Accurate dose calculations in radiotherapy depend on high-quality beam models in the Monaco treatment planning system (TPS). The accelerated go live (AGL) workflow, using a golden beam data (GBD) model, has improved beam modeling accuracy for various linear accelerators (linacs). However, similar studies specifically for the Harmony Pro, recently introduced for online adaptive radiotherapy, have not yet been reported internationally. Moreover, studies on dosimetric differences between TPS beam models with GBD and those with measured beam data (MBD) are limited, and no such studies have been published specifically for Elekta linacs. PURPOSE: This study aimed to assess the clinical performance of GBD in the Monaco TPS for the harmony pro and infinity linacs. METHODS: Beam tuning and data collection were performed on the harmony pro and infinity linacs based on GBD. Subsequently, percentage depth doses (PDDs), off-axis dose profiles, output factors (OFs), absolute doses, and test fields were measured to evaluate the GBD model. Additionally, 31 clinical plans from multiple anatomical sites, including 17 conventional fractionated radiotherapy (CFRT) plans and 14 stereotactic body radiotherapy (SBRT) plans, were designed using the Monaco TPS (GBD model) and practically tested. An Infinity linac with MBD was introduced as a control. RESULTS: PDDs and profiles on both GBD linacs showed 100% passing rate (2%/2 mm). OFs and absolute doses on both GBD linacs agreed within ±1% and ±1.5%, respectively. Additionally, verification of the test fields yielded passing rates above 98% (2%/2 mm) for both GBD linacs. Furthermore, for CFRT plans, measurements on three linacs achieved a passing rate above 95% (3%/2 mm). The absolute dose deviations were within 3%, whereas one MBD linac case exceeded 3% (-3.73%). For SBRT plans, the gamma passing rates were 98.18 ± 1.58%, 98.76 ± 1.54%, and 94.72 ± 0.04% (3%/2 mm) and 96.69 ± 1.96, 96.29 ± 2.26, and 89.51 ± 0.06% (2%/2 mm), for the two GDB linacs and the MDB linac, respectively. The absolute dose deviations were within 3%, whereas two MBD linac cases exceeded 3% (-3.51%, -4.50%). CONCLUSIONS: The harmony pro-GBD and infinity-GBD linac results demonstrated strong agreement with GBD. Clinical plans designed with Monaco TPS (GBD model) were clinically acceptable when delivered on both GBD linacs. Although the test results of GBD model plans delivered on the Infinity-MBD linac showed certain differences compared to those on the two GBD linacs, most plans remained acceptable. This indicates that GBD-based modeling in Monaco TPS offers reliable clinical performance across different linac types.
Assuntos
Neoplasias , Aceleradores de Partículas , Fantomas de Imageamento , Radiocirurgia , Planejamento da Radioterapia Assistida por Computador , Radioterapia de Intensidade Modulada , Planejamento da Radioterapia Assistida por Computador/métodos , Aceleradores de Partículas/instrumentação , Humanos , Dosagem Radioterapêutica , Radioterapia de Intensidade Modulada/métodos , Órgãos em Risco/efeitos da radiação , Neoplasias/radioterapia , Neoplasias/cirurgia , Radiocirurgia/métodos , AlgoritmosRESUMO
Chimeric antigen receptor (CAR) T-cell therapy has revolutionized the treatment of hematologic malignancies, and its application in neuroimmunologic autoimmune diseases is now emerging as a promising therapeutic avenue. Neuroimmunologic diseases such as multiple sclerosis, myasthenia gravis, neuromyelitis optica spectrum disorder, and stiff-person syndrome have shown varying degrees of response to traditional immunomodulatory therapies, but a significant proportion of patients remain refractory to treatment. In recent studies, anti-CD19 CAR T cells have shown encouraging results in targeting B cells, a key driver of autoimmune pathogenesis. CAR T-cells can penetrate the central nervous system and overcome the limitations of conventional B-cell depleting therapies such as rituximab, particularly in accessing ectopic lymphoid follicles that maintain compartmentalized inflammation. In early clinical cases, CAR T-cell treatment has resulted in marked clinical improvements, including significant reductions in symptoms and durable disease remission, with manageable side-effects. In addition, advances in allogeneic CAR T cell constructs and chimeric autoantibody receptor T cells offer additional avenues for precision-targeted therapies. These developments underscore the potential of CAR T cells to reshape the treatment landscape for refractory neuroimmunologic autoimmune diseases and warrant further controlled trials and regulatory exploration.
Assuntos
Algoritmos , Doenças Autoimunes do Sistema Nervoso , Doenças Autoimunes , Imunoterapia Adotiva , Receptores de Antígenos Quiméricos , Linfócitos T , Humanos , Imunoterapia Adotiva/métodos , Imunoterapia Adotiva/tendências , Receptores de Antígenos Quiméricos/imunologia , Doenças Autoimunes do Sistema Nervoso/terapia , Doenças Autoimunes do Sistema Nervoso/imunologia , Linfócitos T/imunologia , Animais , Doenças Autoimunes/terapia , Doenças Autoimunes/imunologiaRESUMO
Editor's Note.-RadioGraphics Update articles supplement or update information found in full-length articles previously published in RadioGraphics. These updates, written by at least one author of the previous article, provide a brief synopsis that emphasizes important new information such as technological advances, revised imaging protocols, new clinical guidelines involving imaging, or updated classification schemes.
Assuntos
Algoritmos , Neoplasias da Mama , Sistemas de Informação em Radiologia , Ultrassonografia Mamária , Feminino , Humanos , Neoplasias da Mama/diagnóstico por imagem , Neoplasias da Mama/classificação , Ultrassonografia Mamária/métodos , Estados UnidosRESUMO
Identifying cancer driver genes is essential for precision oncology, but existing computational methods are often limited by their reliance on single biological networks and their inability to capture long-range molecular dependencies. To address these challenges, we propose GRAFT, a Graph-Aware Fusion Transformer. This framework learns modality-specific features from protein-protein interactions, pathway co-occurrence, and gene semantic similarity using a multi-view graph encoder. These representations are further enriched with two auxiliary feature types: structural encodings derived from network topology and functional embeddings guided by curated gene sets. The integrated features are then processed by a transformer backbone, where a novel edge-attention bias makes the model explicitly sensitive to the underlying graph topologies, enabling the effective modeling of both local and global dependencies. Extensive evaluations demonstrate that GRAFT achieves competitive performance with leading state-of-the-art methods in pan-cancer analysis, while consistently delivering superior predictive accuracy across numerous specific cancer types. More importantly, a functional enrichment analysis of the novel candidate driver genes predicted by our model confirms their strong associations with key cancer-related processes, demonstrating the model's ability to make biologically plausible discoveries. By delivering a powerful and interpretable framework, our model not only advances the identification of cancer driver genes but also establishes a robust paradigm for multimodal data integration in systems biology. The source codes and datasets are publicly accessible at https://github.com/spcho-dev/GRAFT.
Assuntos
Biologia Computacional , Neoplasias , Software , Humanos , Neoplasias/genética , Neoplasias/metabolismo , Biologia Computacional/métodos , Algoritmos , Redes Reguladoras de GenesRESUMO
This work presents the development of an Artificial Intelligence (AI)-based pipeline for patient-specific three-dimensional (3D) reconstruction from oncological magnetic resonance imaging (MRI), leveraging image-derived information to enhance the analysis process. These developments were carried out within the framework of Cella Medical Solutions, forming part of a broader initiative to improve and optimize the company's medical-image processing pipeline. The system integrates automatic MRI sequence classification using a ResNet-based architecture and segmentation of anatomical structures with a modular nnU-Net v2 framework. The classification stage achieved over 90% accuracy and showed improved segmentation performance over prior state-of-the-art pipelines, particularly for contrast-sensitive anatomies such as the hepatic vasculature and pancreas, where dedicated vascular networks showed Dice score differences of approximately 20-22%, and for musculoskeletal structures, where the model outperformed specialized networks in several elements. In terms of computational efficiency, the complete processing of a full MRI case, including sequence classification and segmentation, required approximately four minutes on the target hardware. The integration of sequence-aware information allows for a more comprehensive understanding of MRI signals, leading to more accurate delineations than approaches without such differentiation. From a clinical perspective, the proposed method has the potential to be integrated into surgical planning workflows. The segmentation outputs were converted into a patient-specific 3D model, which was subsequently integrated into Cella's surgical planner as a proof of concept. This process illustrates the transition from voxel-wise anatomical labels to a fully navigable 3D reconstruction, representing a step toward more robust and personalized AI-driven medical-image analysis workflows that leverage sequence-aware information for enhanced clinical utility.
Assuntos
Inteligência Artificial , Processamento de Imagem Assistido por Computador , Imageamento por Ressonância Magnética , Neoplasias , Humanos , Imageamento por Ressonância Magnética/métodos , Imageamento Tridimensional/métodos , Neoplasias/diagnóstico por imagem , Neoplasias/cirurgia , Processamento de Imagem Assistido por Computador/métodos , AlgoritmosRESUMO
Importance: Over 1 million pulmonary nodules are discovered each year in the US, and many of these undergo molecular imaging-guided surgery to obtain a diagnosis. Locating a small nodule and determining its malignant potential is technically challenging and is prone to human error. Objective: To demonstrate use of a machine learning (ML) algorithm with molecular imaging to analyze imaging data during lung cancer surgery to determine malignant potential of nodules. Design, Setting, and Participants: Data were retrospectively analyzed from a prospectively collected database. Between 2014 and 2021, patients at the hospital of the University of Pennsylvania with lung nodules were included in the study. Patients in the model development set were randomly allocated into training and validation sets in an 8:2 ratio. Data were analyzed from January 2014 and December 2021. Main Outcomes and Measures: Algorithmic tumor to background ratio (TBR) detection was implemented for individual images using Image Processing Toolkit. Developed nomogram and artificial intelligence (AI) image analyzer were combined as an optical biopsy algorithm and tested prospectively between 2021 and 2024. Results: A total of 322 patients with lung nodules were included in the study, of whom 279 had complete clinical data for data analysis (175 [62.7%] female). The nomograms and image segmentation technology were developed using a large database of IMI videos (1014 video sequences) and demonstrated an area under the curve of 0.865 to 0.893 for malignant nodule assessment. On multivariate logistic regression analysis, patient smoking history of greater than 5 pack-years (patient pack-years [PPY] >5), ex vivo back table TBR greater than 2.0, ex vivo bisected tumor lesions TBR greater than 2.4, and in situ (inside the chest) fluorescence were found to have statistically significant associations with malignancy on final pathology. Prospective testing in an independent set of 61 consecutive patients during IMI-guided cancer surgery demonstrated a sensitivity of 93.8%, specificity of 100%, positive predictive value of 100%, and negative predictive value of 71%. The study algorithm determined malignant potential of the lesion in less than 2 minutes (mean [SD], 1.8 [0.17] minutes) compared with a mean (SD) of 34 (11) minutes with frozen section analysis. Conclusion: In this cohort study of patients with indeterminate lung nodules, intraoperative imaging data analyzed by AI accurately determined if a nodule was malignant. This has the potential to improve the diagnostic challenges that occur at the time of surgery.
Assuntos
Neoplasias Pulmonares , Aprendizado de Máquina , Imageamento Molecular , Nódulos Pulmonares Múltiplos , Nódulo Pulmonar Solitário , Cirurgia Assistida por Computador , Humanos , Neoplasias Pulmonares/cirurgia , Neoplasias Pulmonares/diagnóstico por imagem , Neoplasias Pulmonares/patologia , Neoplasias Pulmonares/diagnóstico , Feminino , Masculino , Pessoa de Meia-Idade , Idoso , Estudos Retrospectivos , Imageamento Molecular/métodos , Cirurgia Assistida por Computador/métodos , Nódulos Pulmonares Múltiplos/diagnóstico por imagem , Nódulos Pulmonares Múltiplos/cirurgia , Algoritmos , Nódulo Pulmonar Solitário/diagnóstico por imagem , Nódulo Pulmonar Solitário/cirurgia , Nódulo Pulmonar Solitário/patologiaRESUMO
MOTIVATION: The interaction between peptides and human leukocyte antigen class II (HLA-II) molecules plays a pivotal role in adaptive immune responses, as HLA-II mediates the recognition of exogenous antigens and initiates T cell activation through peptide presentation. Accurate prediction of peptide-HLA-II binding serves as a cornerstone for deciphering cellular immune responses, and is essential for guiding the optimization of antibody therapeutics. Researchers have developed several computational approaches to identify peptide-HLA-II interaction and presentation. However, most computational approaches exhibit inconsistent predictive performance, poor generalization ability and limited biological interpretability. RESULTS: In this study, we present DSCA-HLAII, a novel predictive framework for peptide-HLA-II interactions and presentation based on a dual-stream cross-attention architecture. The framework proposes a dual-stream cross-attention (DSCA) mechanism to integrate pre-trained semantic embedding ESMC with sequence-level ONE-HOT features. The DSCA mechanism effectively models the interaction dynamics between peptides and HLA-II molecules, enabling the precise identification of key binding sites. Experimental results demonstrate that DSCA-HLAII consistently surpasses existing state-of-the-art approaches, demonstrating high accuracy and robustness in predicting peptide-HLA-II interactions and presentation. We further demonstrate the capability of DSCA-HLAII for predicting peptide binding cores and assessing antibody immunogenicity, which is expected to advance artificial intelligence-based peptide drug discovery.
Assuntos
Apresentação de Antígeno , Biologia Computacional , Antígenos de Histocompatibilidade Classe II , Peptídeos , Humanos , Peptídeos/química , Peptídeos/metabolismo , Peptídeos/imunologia , Biologia Computacional/métodos , Ligação Proteica , Antígenos de Histocompatibilidade Classe II/química , Antígenos de Histocompatibilidade Classe II/metabolismo , Antígenos de Histocompatibilidade Classe II/imunologia , Sítios de Ligação , Apresentação de Antígeno/imunologia , AlgoritmosRESUMO
Endometrial cancer (EC) is the most common gynecological malignancy, yet reliable screening and diagnostic approaches remain limited. We developed a weakly supervised deep multi-instance learning model (DSMIL) to classify hematoxylin and eosin-stained whole-slide images (WSIs) of endometrial tissue. A total of 885 WSIs from 442 patients, including EC, atypical endometrial hyperplasia (AEH), endometrial hyperplasia without atypia (EH), and normal endometrium (NE), were analyzed. DSMIL achieved an average AUROC of 0.9776 for four-class classification, with inter-class AUROCs of 0.9876 for EC, 0.9600 for AEH, 0.9771 for EH, and 0.9855 for NE, and outperformed other algorithms such as TransMSL, CLAM, and ABMIL (average accuracy = 0.8914). Attention heatmaps highlighted regions associated with pathological features, while nnU-Net v2 combined with HoverNet enabled identification of atypical glandular epithelial cells, which showed increased density, size, and perimeter but reduced axis ratios compared with normal cells. These results suggest that DSMIL provides a reliable computational pathology approach for the classification of endometrial lesions and the characterization of atypical cells.
Assuntos
Aprendizagem Profunda , Hiperplasia Endometrial , Neoplasias do Endométrio , Endométrio , Feminino , Humanos , Hematoxilina/química , Eosina Amarelada/química , Endométrio/patologia , Neoplasias do Endométrio/patologia , Neoplasias do Endométrio/classificação , Neoplasias do Endométrio/diagnóstico , Hiperplasia Endometrial/patologia , Hiperplasia Endometrial/classificação , Hiperplasia Endometrial/diagnóstico , Algoritmos , Coloração e RotulagemRESUMO
BACKGROUND: Progress in mitigating plan degradation due to intrafraction patient motion may involve the identification and management of specific control points that are sensitive to motion. Robust planning in this manner could improve deliverable dosimetry and support advancements toward reducing planning target volume (PTV) margins. PURPOSE: To improve radiotherapy plan quality robustness in the presence of intrafraction motion by identifying the control-point-specific dosimetric sensitivities. This work explores control-point-specific plan characteristics that impact dosimetry by retrospectively assessing the consequence of simulated patient scenarios for cranial radiotherapy. METHODS: Single target cranial volumetric modulated arc therapy (VMAT) treatment plans (n = 30) were converted into static field plans and reconstructed by applying 3D control-point-specific motion traces (n = 100) using our in-house MATLAB application. PTV coverage (volume covered by 100% of the prescription isodose, VRx) and the differences in minimum dose delivered to 99% (D99%) of the gross tumor volume (GTV) were examined across the patient cohort as these are pertinent metrics for each structure. To identify the individual control points where motion led to target coverage loss, three patient plans (5 and 14 were randomly chosen, and 19 with the greatest range in prescription dose coverage) were selected for an area under the curve (AUC) analysis of control point dose volume histograms (DVHs). The mean dose difference in the area under the curve of control point DVHs (mAUC), and the standard deviation of differences (sAUC) were the metrics used in the investigation. Multileaf collimator (MLC) aperture areas were also explored as a function of these metrics. RESULTS: Under conditions of simulated intrafraction motion, PTV coverage spanned from -2.8% to +0.73% of target volume with 78.6% of the three thousand motion traces resulting in coverage loss. There were no changes in GTV D99% that exceeded ± 1.5%. For the in-depth control point analysis, MLC aperture areas formed weak to moderately weak correlations with sAUC (r = -0.19, r = -0.42, and r = -0.32, p < 0.01 for patient plans 5, 14, and 19 respectively). In addition, two statistically distinct sub-populations of MLC aperture areas were confirmed by Welch corrected t-tests (p < 0.0001, p = 0.02, p = 0.005 for cases 5, 14 and 19) across a threshold of ± 0.05 mGy in mAUC. CONCLUSION: This work has demonstrated that the dosimetric impact of intrafractional motion reflects the inherent motion sensitivity of specific control points. Our findings suggest that motion sensitive control points could be selectively targeted for gating to enhance robustness against intrafraction motion and improve dosimetry in support of a PTV margin reduction strategy. Single target cranial plans serve as ideal cases to characterize the consequences of motion at the control point level with the aim of expanding the analysis to other anatomical regions.
Assuntos
Neoplasias Encefálicas , Irradiação Craniana , Órgãos em Risco , Planejamento da Radioterapia Assistida por Computador , Radioterapia de Intensidade Modulada , Humanos , Planejamento da Radioterapia Assistida por Computador/métodos , Planejamento da Radioterapia Assistida por Computador/normas , Radioterapia de Intensidade Modulada/métodos , Dosagem Radioterapêutica , Órgãos em Risco/efeitos da radiação , Estudos Retrospectivos , Neoplasias Encefálicas/radioterapia , Algoritmos , Irradiação Craniana/métodosRESUMO
PURPOSE: In this study, a bias field correction workflow was proposed to improve the flexibility and generalizability of the generative adversarial network (GAN) model for abdominal cancer patients treated with a 0.35T magnetic resonance imaging linear accelerator (MR-LINAC) system. METHODS: Model training was performed using brain MR images acquired on a 3T diagnostic scanner, while model testing was performed using abdominal MR images obtained using a 0.35T MR-LINAC system. The performance of the proposed workflow was first compared with the GAN model using root-mean-square error (RMSE), peak signal-to-noise ratio (PSNR), and structural similarity index measure (SSIM). To assess the impact of the workflow on image segmentation, it was also compared with the N4ITK algorithm. Segmentation was performed using the k-means clustering algorithm with three clusters corresponding to air, fat, and soft tissue. Segmentation accuracy was then evaluated using the Dice similarity coefficient (DSC). RESULTS: The RMSE values were 30.59, 12.06, 10.37 for the bias field-corrupted images (IIN), GAN-corrected images (IGAN), and images corrected with the proposed workflow (IOUT), respectively. Corresponding PSNR values were 42.34, 46.04, 47.04 dB, and SSIM values were 0.84, 0.96, 0.98. For segmentation accuracy, the mean DSC for air masks was 0.95, 0.97, and 0.97; for fat masks, 0.61, 0.71, and 0.74; and for soft tissue masks, 0.60, 0.68, and 0.69, corresponding to IIN, N4ITK-corrected images (IN4ITK), and IOUT, respectively CONCLUSION: By effectively mitigating bias field artifacts, the proposed workflow has the potential to strengthen the clinical utility of MRI-guided adaptive radiotherapy for abdominal cancers, ensuring safer and more accurate radiation delivery.
Assuntos
Neoplasias Abdominais , Algoritmos , Processamento de Imagem Assistido por Computador , Imageamento por Ressonância Magnética , Redes Neurais de Computação , Aceleradores de Partículas , Planejamento da Radioterapia Assistida por Computador , Fluxo de Trabalho , Humanos , Neoplasias Abdominais/radioterapia , Neoplasias Abdominais/diagnóstico por imagem , Imageamento por Ressonância Magnética/métodos , Processamento de Imagem Assistido por Computador/métodos , Planejamento da Radioterapia Assistida por Computador/métodos , Aceleradores de Partículas/instrumentação , Dosagem Radioterapêutica , Radioterapia de Intensidade Modulada/métodos , Redes Generativas AdversariaisRESUMO
PURPOSE: To develop clinically relevant test cases for commissioning Model-Based Dose Calculation Algorithms (MBDCAs) for 192Ir High Dose Rate (HDR) gynecologic brachytherapy following the workflow proposed by the TG-186 report and the WGDCAB report 372. ACQUISITION AND VALIDATION METHODS: Two cervical cancer intracavitary HDR brachytherapy models were developed based on a real patient, using either uniformly structured regions or realistic segmentation. The patient's computed tomography (CT) images were processed, converted to a series of digital imaging and communications in medicine (DICOM) CT images, and imported into two treatment planning systems (TPSs), the Oncentra Brachy and BrachyVision. The original segmentation of the clinical case was augmented to enable a thorough dosimetric analysis. The actual clinical treatment plan was generally maintained, with the source replaced by a generic 192Ir HDR source. Dose to medium in medium calculations were performed using the MBDCA option of each TPS, and three different Monte Carlo (MC) simulation codes. MC results demonstrated agreement within statistical uncertainty, while comparisons between the commercial TPS MBDCAs and a general-purpose MC code highlighted both the advantages and limitations of the studied MBDCAs, suggesting potential approaches to overcome the challenges. DATA FORMAT AND USAGE NOTES: The datasets for the developed cases are available online at https://doi.org/10.5281/zenodo.15720996. The DICOM files include the treatment plan for each case, TPS, and the corresponding reference MC dose data. The package also contains a TPS- and case-specific user guide for commissioning the MBDCAs, as well as files necessary to replicate the MC simulations. POTENTIAL APPLICATIONS: The provided datasets and proposed methodology can serve as a commissioning framework for TPSs that employ MBDCAs, as well as a benchmark for brachytherapy researchers using MC methods and MBDCA developers. They also facilitate intercomparisons of MBDCA performance and provide a quality assurance resource for evaluating future TPS software updates.
Assuntos
Algoritmos , Braquiterapia , Radioisótopos de Irídio , Garantia da Qualidade dos Cuidados de Saúde , Doses de Radiação , Planejamento da Radioterapia Assistida por Computador , Neoplasias do Colo do Útero , Braquiterapia/métodos , Braquiterapia/instrumentação , Humanos , Radioisótopos de Irídio/uso terapêutico , Dosagem Radioterapêutica , Planejamento da Radioterapia Assistida por Computador/métodos , Feminino , Benchmarking , Método de Monte Carlo , Neoplasias do Colo do Útero/radioterapia , Neoplasias do Colo do Útero/diagnóstico por imagem , Controle de QualidadeRESUMO
BACKGROUND: Digital breast tomosynthesis (DBT) has become standard practice; however, the acquisition method of DBT between vendors is far from standardized. Currently, there are three commercially available DBT tube motion techniques: (1) continuous motion, (2) step-and-shoot, and (3) continuous motion with flying focal spot. Each of these methods represents a trade-off between total acquisition time and focal spot blur. PURPOSE: The aim of the study was to characterize the increase in effective focal spot size in DBT relative to standard 2D projections and assess the influence of this increase on spatial resolution using the modulation transfer function (MTF). METHODS: Focal spot size was measured for both a 2D acquisition and the 0° DBT projection using a 10 µm slit phantom. Imaging techniques were set to those used for a 2, 4, and 8 cm thick breast of 50/50 adipose/fat composition. MTF curves were measured using a copper edge phantom both at the breast support plane and 4 cm above the breast support. RESULTS: The effective focal spot size increase from 2D to DBT increased with breast thickness for all systems. The continuous motion systems showed the greatest increase in effective focal spot size with percent increases of 101% to 462% depending on unit and breast thickness. The flying focal spot system showed the smallest increase in effective focal spot size in DBT acquisitions, being 3%, 21%, and 25% for a 2, 4, and 8 cm thick breast, respectively. The step-and-shoot and flying focal spot systems showed no degradation in MTF curves due to increasing effective focal spot size in DBT acquisitions, while the continuous motion systems showed a reduction in the frequency at which the MTF curve reached 50% of 26%-45%. CONCLUSION: Both step-and-shoot and flying focal spot systems minimized effective focal spot size increase in DBT acquisitions compared to continuous tube motion systems.