Hasil untuk "Pathology"

Vida Adeli, Soroush Mehraban, Majid Mirmehdi et al.

Gait analysis is crucial for the diagnosis and monitoring of movement disorders like Parkinson's Disease. While computer vision models have shown potential for objectively evaluating parkinsonian gait, their effectiveness is limited by scarce clinical datasets and the challenge of collecting large and well-labelled data, impacting model accuracy and risk of bias. To address these gaps, we propose GAITGen, a novel framework that generates realistic gait sequences conditioned on specified pathology severity levels. GAITGen employs a Conditional Residual Vector Quantized Variational Autoencoder to learn disentangled representations of motion dynamics and pathology-specific factors, coupled with Mask and Residual Transformers for conditioned sequence generation. GAITGen generates realistic, diverse gait sequences across severity levels, enriching datasets and enabling large-scale model training in parkinsonian gait analysis. Experiments on our new PD-GaM (real) dataset demonstrate that GAITGen outperforms adapted state-of-the-art models in both reconstruction fidelity and generation quality, accurately capturing critical pathology-specific gait features. A clinical user study confirms the realism and clinical relevance of our generated sequences. Moreover, incorporating GAITGen-generated data into downstream tasks improves parkinsonian gait severity estimation, highlighting its potential for advancing clinical gait analysis.

Matthew Sotoudeh

Although existing garbage collectors (GCs) perform extremely well on typical programs, there still exist pathological programs for which modern GCs significantly degrade performance. This observation begs the question: might there exist a 'holy grail' GC algorithm, as yet undiscovered, guaranteeing both constant-length pause times and that memory is collected promptly upon becoming unreachable? For decades, researchers have understood that such a GC is not always possible, i.e., some pathological behavior is unavoidable when the program can make heap cycles and operates near the memory limit, regardless of the GC algorithm used. However, this understanding has until now been only informal, lacking a rigorous formal proof. This paper complements that informal understanding with a rigorous proof, showing with mathematical certainty that every GC that can implement a realistic mutator-observer interface has some pathological program that forces it to either introduce a long GC pause into program execution or reject an allocation even though there is available space. Hence, language designers must either accept these pathological scenarios and design heuristic approaches that minimize their impact (e.g., generational collection), or restrict programs and environments to a strict subset of the behaviors allowed by our mutator-observer-style interface (e.g., by enforcing a type system that disallows cycles or overprovisioning memory).

Yuxuan Sun, Yixuan Si, Chenglu Zhu et al.

Recent advances in computational pathology have led to the emergence of numerous foundation models. These models typically rely on general-purpose encoders with multi-instance learning for whole slide image (WSI) classification or apply multimodal approaches to generate reports directly from images. However, these models cannot emulate the diagnostic approach of pathologists, who systematically examine slides at low magnification to obtain an overview before progressively zooming in on suspicious regions to formulate comprehensive diagnoses. Instead, existing models directly output final diagnoses without revealing the underlying reasoning process. To address this gap, we introduce CPathAgent, an innovative agent-based approach that mimics pathologists' diagnostic workflow by autonomously navigating across WSI based on observed visual features, thereby generating substantially more transparent and interpretable diagnostic summaries. To achieve this, we develop a multi-stage training strategy that unifies patch-level, region-level, and WSI-level capabilities within a single model, which is essential for replicating how pathologists understand and reason across diverse image scales. Additionally, we construct PathMMU-HR2, the first expert-validated benchmark for large region analysis. This represents a critical intermediate scale between patches and whole slides, reflecting a key clinical reality where pathologists typically examine several key large regions rather than entire slides at once. Extensive experiments demonstrate that CPathAgent consistently outperforms existing approaches across benchmarks at three different image scales, validating the effectiveness of our agent-based diagnostic approach and highlighting a promising direction for computational pathology.

Kazuya Nishimura, Haruka Hirose, Ryoma Bise et al.

Gene expression estimation from pathology images has the potential to reduce the RNA sequencing cost. Point-wise loss functions have been widely used to minimize the discrepancy between predicted and absolute gene expression values. However, due to the complexity of the sequencing techniques and intrinsic variability across cells, the observed gene expression contains stochastic noise and batch effects, and estimating the absolute expression values accurately remains a significant challenge. To mitigate this, we propose a novel objective of learning relative expression patterns rather than absolute levels. We assume that the relative expression levels of genes exhibit consistent patterns across independent experiments, even when absolute expression values are affected by batch effects and stochastic noise in tissue samples. Based on the assumption, we model the relation and propose a novel loss function called STRank that is robust to noise and batch effects. Experiments using synthetic datasets and real datasets demonstrate the effectiveness of the proposed method. The code is available at https://github.com/naivete5656/STRank.

Mahdi Amiri, Hatef Otroshi Shahreza, Ina Kodrasi

Automatic pathological speech detection approaches have shown promising results, gaining attention as potential diagnostic tools alongside costly traditional methods. While these approaches can achieve high accuracy, their lack of interpretability limits their applicability in clinical practice. In this paper, we investigate the use of multimodal Large Language Models (LLMs), specifically ChatGPT-4o, for automatic pathological speech detection in a few-shot in-context learning setting. Experimental results show that this approach not only delivers promising performance but also provides explanations for its decisions, enhancing model interpretability. To further understand its effectiveness, we conduct an ablation study to analyze the impact of different factors, such as input type and system prompts, on the final results. Our findings highlight the potential of multimodal LLMs for further exploration and advancement in automatic pathological speech detection.

Mahesh Sultania, Priyansh Jain, Itisha Chaudhary et al.

Objective The preoperative (clinical and radiological) depth of invasion (DOI) in early tumours can guide the surgeons in deciding elective neck dissection, the extent of invasive surgery, the need for reconstruction, potential adjuvant treatment and the patient's prognosis. This study aimed to validate the 8th AJCC clinical T stage using DOI, assess interobserver bias, and correlate the clinical, radiological and pathological T stages. Materials and methods: This was a prospective clinical study carried out from December 2019 to June 2023 at an academic tertiary care centre. Patients with squamous cell carcinoma of oral cavity and lip without involvement of skin or bone undergoing upfront surgery were included. The clinical assessment of T stage using DOI was done by three examiners, blinded to each other. The radiological T stage was assessed by an MRI and pathological T stage on fixed formalin surgical specimen. Results: A total of 173 patients were assessed during the study period, out of which 44 met the inclusion criteria. There was a fair agreement regarding the combined clinical T stage between all three examines. A statistically significant correlation was found between the clinical and pathological T stage (p-value 0.010), the radiological and pathological T stage (p-value 0.001) and clinical and radiological T stage (p-value 0.004). Conclusion: The clinical and radiological T stage using DOI correlated well with the pathological T stage. The 8th AJCC clinical T stage of oral squamous cell carcinoma was accurate in our study population.

Oana Baldasici, Olga Soritau, Andrei Roman et al.

Abstract Background Cancer stem-like cells (CSCs) have been extensively researched as the primary drivers of therapy resistance and tumor relapse in patients with breast cancer. However, due to lack of specific molecular markers, increased phenotypic plasticity and no clear clinicopathological features, the assessment of CSCs presence and functionality in solid tumors is challenging. While several potential markers, such as CD24/CD44, have been proposed, the extent to which they truly represent the stem cell potential of tumors or merely provide static snapshots is still a subject of controversy. Recent studies have highlighted the crucial role of the tumor microenvironment (TME) in influencing the CSC phenotype in breast cancer. The interplay between the tumor and TME induces significant changes in the cancer cell phenotype, leading to the acquisition of CSC characteristics, therapeutic resistance, and metastatic spread. Simultaneously, CSCs actively shape their microenvironment by evading immune surveillance and attracting stromal cells that support tumor progression. Methods In this study, we associated in vitro mammosphere formation assays with bulk tumor microarray profiling and deconvolution algorithms to map CSC functionality and the microenvironmental landscape in a large cohort of 125 breast tumors. Results We found that the TME score was a significant factor associated with CSC functionality. CSC-rich tumors were characterized by an immune-suppressed TME, while tumors devoid of CSC potential exhibited high immune infiltration and activation of pathways involved in the immune response. Gene expression analysis revealed IFNG, CXCR5, CD40LG, TBX21 and IL2RG to be associated with the CSC phenotype and also displayed prognostic value for patients with breast cancer. Conclusion These results suggest that the characterization of CSCs content and functionality in tumors can be used as an attractive strategy to fine-tune treatments and guide clinical decisions to improve patients therapy response.

Ivo Dilber, Stjepko Pleština, Stjepko Pleština et al.

Gastric cancer ranks fourth among the most commonly diagnosed cancers, with over a million new cases diagnosed worldwide each year. Acute and chronic kidney damage are common in patients with malignant diseases and are associated with increased risk of complications and mortality. Rarely, acute renal insufficiency may result from bilateral infiltration of renal parenchyma by tumor cells from another organ. We present a case of a patient with clinical suspected gastric cancer and metastases to the kidneys leading to acute renal failure requiring hemodialysis. Despite gastric biopsies, no tumor cells were found, while histopathological examination of enlarged intra-abdominal lymph node biopsy material confirmed adenocarcinoma of signet ring cell originating from the digestive system. Stomach cancer was identified as the most likely primary site after the kidney biopsy was performed. To the best of our knowledge, no case of gastric cancer leading to kidney metastases and acute renal failure requiring renal replacement therapy was yet described. Multidisciplinary collaboration among oncologists, urologists, radiologists, pathologists, and nephrologists is essential for the optimal treatment outcome of these patients, who generally have a poor prognosis.

Shu Yang, Yihui Wang, Hao Chen

Multiple Instance Learning (MIL) has emerged as a dominant paradigm to extract discriminative feature representations within Whole Slide Images (WSIs) in computational pathology. Despite driving notable progress, existing MIL approaches suffer from limitations in facilitating comprehensive and efficient interactions among instances, as well as challenges related to time-consuming computations and overfitting. In this paper, we incorporate the Selective Scan Space State Sequential Model (Mamba) in Multiple Instance Learning (MIL) for long sequence modeling with linear complexity, termed as MambaMIL. By inheriting the capability of vanilla Mamba, MambaMIL demonstrates the ability to comprehensively understand and perceive long sequences of instances. Furthermore, we propose the Sequence Reordering Mamba (SR-Mamba) aware of the order and distribution of instances, which exploits the inherent valuable information embedded within the long sequences. With the SR-Mamba as the core component, MambaMIL can effectively capture more discriminative features and mitigate the challenges associated with overfitting and high computational overhead. Extensive experiments on two public challenging tasks across nine diverse datasets demonstrate that our proposed framework performs favorably against state-of-the-art MIL methods. The code is released at https://github.com/isyangshu/MambaMIL.

Abhijeet Patil, Harsh Diwakar, Jay Sawant et al.

Histopathology whole slide images (WSIs) are being widely used to develop deep learning-based diagnostic solutions, especially for precision oncology. Most of these diagnostic softwares are vulnerable to biases and impurities in the training and test data which can lead to inaccurate diagnoses. For instance, WSIs contain multiple types of tissue regions, at least some of which might not be relevant to the diagnosis. We introduce HistoROI, a robust yet lightweight deep learning-based classifier to segregate WSI into six broad tissue regions -- epithelium, stroma, lymphocytes, adipose, artifacts, and miscellaneous. HistoROI is trained using a novel human-in-the-loop and active learning paradigm that ensures variations in training data for labeling-efficient generalization. HistoROI consistently performs well across multiple organs, despite being trained on only a single dataset, demonstrating strong generalization. Further, we have examined the utility of HistoROI in improving the performance of downstream deep learning-based tasks using the CAMELYON breast cancer lymph node and TCGA lung cancer datasets. For the former dataset, the area under the receiver operating characteristic curve (AUC) for metastasis versus normal tissue of a neural network trained using weakly supervised learning increased from 0.88 to 0.92 by filtering the data using HistoROI. Similarly, the AUC increased from 0.88 to 0.93 for the classification between adenocarcinoma and squamous cell carcinoma on the lung cancer dataset. We also found that the performance of the HistoROI improves upon HistoQC for artifact detection on a test dataset of 93 annotated WSIs. The limitations of the proposed model are analyzed, and potential extensions are also discussed.

Sujin Park, Junsik Kim, Woncheol Jang et al.

Background Gallbladder cancer (GBC) is usually detected in advanced stages with a low 5-year survival rate. Delta-like ligand 4 (DLL4), vascular endothelial growth factor (VEGF), and hypoxia-inducible factor-2alpha (HIF2α) have been studied for their role in tumorigenesis and potential for therapeutic target, and multiple clinical trials of the agents targeting them are ongoing. We investigated the expression of these markers in surgically resected GBC and tried to reveal their association with the clinicopathologic features, mutual correlation of their expression, and prognosis of the GBC patients by their expression. Methods We constructed the tissue microarray blocks of 99 surgically resected GBC specimens and performed immunohistochemistry of DLL4, VEGF, and HIF2α. We used the quantitative digital image analysis to evaluate DLL4 and VEGF expression, while the expression of HIF2α was scored manually. Results The expression of VEGF and HIF2α showed a significant trend with tumor differentiation (p = .028 and p = .006, respectively). We found that the high DLL4 and VEGF expression were significantly correlated with lymph node metastasis (p = .047, both). The expression of VEGF and HIF2α were significantly correlated (p < .001). The GBC patients with low HIF2α expression showed shorter recurrence-free survival than those with high HIF2α expression. Conclusions This study suggested the possibility of the usage of DLL4 and VEGF to predict the lymph node metastasis and the possibility of VEGF and HIF2α to predict the expression level mutually. Further studies may be needed to validate our study results and eventually accelerate the introduction of the targeted therapy in GBC.

Arnault Tauziède-Espariat, Emmanuelle Uro-Coste, Yvan Nicaise et al.

Trudi de Koning, Huib Versnel, Joost Meekes et al.

Achtergrond en doel Kinderen met epilepsie scoren bij taalonderzoek meestal significant lager dan gezonde leeftijdgenoten. Tot beter begrip hiervan vergeleken wij kinderen met farmacoresistente epilepsie met gezonde leeftijdgenoten in een benoemtaak als index van het lexicon. Onderzoeksvragen waren: verschillen de groepen in aantallen goede benoemingen en/of in hoeveelheid baat bij geboden hulp; hebben verwervingsleeftijd en gebruiksfrequentie van de woorden (namen van de afgebeelde voorwerpen) invloed; zijn de verschillen door de kindertijd heen gelijk; en is invloed vaststelbaar van epilepsievariabelen, demografische variabelen (leeftijd, geslacht, intelligentiequotiënt), en van de omgevingsvariabele opleiding van de ouders. Methode Vijfenveertig kinderen met farmacoresistente epilepsie (leeftijd 3,4 - 17,9 jaar; 20 meisjes) en 86 gezonde, per patiënt op geslacht en leeftijd gematchte kinderen/jongeren benoemden lijntekeningen van voorwerpen die verwezen naar vroeggeleerde hoogfrequente (VH), vroeggeleerde laagfrequente (VL) en laatgeleerde laagfrequente woorden (LL). Bij een willekeurig deel van de onjuiste of uitblijvende benoemingen gaf de onderzoeker hulp door een vraag te stellen naar de functie van het afgebeelde voorwerp. Als dat niet hielp, volgde een fonologische aanwijzing. Goede benoemingen en baat bij de hulp werden per categorie woorden geanalyseerd. Groepsverschillen werden niet-parametrisch getoetst. Exploratief werden met lineaire regressie ziekte- en andere invloeden geanalyseerd. Resultaten Er was geen significant verschil in de percentages epilepsie- en controlekinderen die direct alle plaatjes goed benoemden. Ook de percentages kinderen die baat hadden bij hulp verschilden niet significant tussen beide groepen. Wel waren de gemiddelde benoemscores van de kinderen met epilepsie lager dan die van de controlekinderen. Dit groepsverschil verdween met toenemende leeftijd in de VH- en VL-woorden maar niet in de LL-woorden. De baat bij hulp nam in beide groepen toe met de leeftijd als het ging om VH- en VL-woorden. Voor LL-woorden toonde de epilepsiegroep minder leeftijdgebonden toename. Hoe vroeger in het leven de epilepsie was ontstaan, des te zwakker was het benoemen van de plaatjes die verwezen naar LL-woorden. Conclusie Kinderen met farmacoresistente epilepsie kunnen hun achterstand in ontwikkeling van het lexicon althans voor vroeg geleerde woorden inhalen. Ook bij kinderen met epilepsie stimuleert hulp (i.c. vragen naar de functie van het voorwerp of een fonologische aanwijzing) het zoeken in het lexicon.

Sonia Brahim, Ana-Maria Negulescu, Clara Geneste et al.

Abstract Notch signaling is a conserved signaling pathway that participates in many aspects of mammary gland development and homeostasis, and has extensively been associated with breast tumorigenesis. Here, to unravel the as yet debated role of Notch3 in breast cancer development, we investigated its expression in human breast cancer samples and effects of its loss in mice. Notch3 expression was very weak in breast cancer cells and was associated with good patient prognosis. Interestingly, its expression was very strong in stromal cells of these patients, though this had no prognostic value. Mechanistically, we demonstrated that Notch3 prevents tumor initiation via HeyL-mediated inhibition of Mybl2, an important regulator of cell cycle. In the mammary glands of Notch3-deficient mice, we observed accelerated tumor initiation and proliferation in a MMTV-Neu model. Notch3-null tumors were enriched in Mybl2 mRNA signature and protein expression. Hence, our study reinforces the anti-tumoral role of Notch3 in breast tumorigenesis.

Daniel du Plessis

Wentao Zhu, Yuan Jin, Gege Ma et al.

The accurate diagnosis on pathological subtypes for lung cancer is of significant importance for the follow-up treatments and prognosis managements. In this paper, we propose self-generating hybrid feature network (SGHF-Net) for accurately classifying lung cancer subtypes on computed tomography (CT) images. Inspired by studies stating that cross-scale associations exist in the image patterns between the same case's CT images and its pathological images, we innovatively developed a pathological feature synthetic module (PFSM), which quantitatively maps cross-modality associations through deep neural networks, to derive the "gold standard" information contained in the corresponding pathological images from CT images. Additionally, we designed a radiological feature extraction module (RFEM) to directly acquire CT image information and integrated it with the pathological priors under an effective feature fusion framework, enabling the entire classification model to generate more indicative and specific pathologically related features and eventually output more accurate predictions. The superiority of the proposed model lies in its ability to self-generate hybrid features that contain multi-modality image information based on a single-modality input. To evaluate the effectiveness, adaptability, and generalization ability of our model, we performed extensive experiments on a large-scale multi-center dataset (i.e., 829 cases from three hospitals) to compare our model and a series of state-of-the-art (SOTA) classification models. The experimental results demonstrated the superiority of our model for lung cancer subtypes classification with significant accuracy improvements in terms of accuracy (ACC), area under the curve (AUC), and F1 score.

Leo Fillioux, Joseph Boyd, Maria Vakalopoulou et al.

Multiple instance learning is an ideal mode of analysis for histopathology data, where vast whole slide images are typically annotated with a single global label. In such cases, a whole slide image is modelled as a collection of tissue patches to be aggregated and classified. Common models for performing this classification include recurrent neural networks and transformers. Although powerful compression algorithms, such as deep pre-trained neural networks, are used to reduce the dimensionality of each patch, the sequences arising from whole slide images remain excessively long, routinely containing tens of thousands of patches. Structured state space models are an emerging alternative for sequence modelling, specifically designed for the efficient modelling of long sequences. These models invoke an optimal projection of an input sequence into memory units that compress the entire sequence. In this paper, we propose the use of state space models as a multiple instance learner to a variety of problems in digital pathology. Across experiments in metastasis detection, cancer subtyping, mutation classification, and multitask learning, we demonstrate the competitiveness of this new class of models with existing state of the art approaches. Our code is available at https://github.com/MICS-Lab/s4_digital_pathology.

Shaheer U. Saeed, Tom Syer, Wen Yan et al.

We propose an image synthesis mechanism for multi-sequence prostate MR images conditioned on text, to control lesion presence and sequence, as well as to generate paired bi-parametric images conditioned on images e.g. for generating diffusion-weighted MR from T2-weighted MR for paired data, which are two challenging tasks in pathological image synthesis. Our proposed mechanism utilises and builds upon the recent stable diffusion model by proposing image-based conditioning for paired data generation. We validate our method using 2D image slices from real suspected prostate cancer patients. The realism of the synthesised images is validated by means of a blind expert evaluation for identifying real versus fake images, where a radiologist with 4 years experience reading urological MR only achieves 59.4% accuracy across all tested sequences (where chance is 50%). For the first time, we evaluate the realism of the generated pathology by blind expert identification of the presence of suspected lesions, where we find that the clinician performs similarly for both real and synthesised images, with a 2.9 percentage point difference in lesion identification accuracy between real and synthesised images, demonstrating the potentials in radiological training purposes. Furthermore, we also show that a machine learning model, trained for lesion identification, shows better performance (76.2% vs 70.4%, statistically significant improvement) when trained with real data augmented by synthesised data as opposed to training with only real images, demonstrating usefulness for model training.

Yang Liu, Shi Gu

The registration of pathological images plays an important role in medical applications. Despite its significance, most researchers in this field primarily focus on the registration of normal tissue into normal tissue. The negative impact of focal tissue, such as the loss of spatial correspondence information and the abnormal distortion of tissue, are rarely considered. In this paper, we propose GIRNet, a novel unsupervised approach for pathological image registration by incorporating segmentation and inpainting through the principles of Generation, Inpainting, and Registration (GIR). The registration, segmentation, and inpainting modules are trained simultaneously in a co-learning manner so that the segmentation of the focal area and the registration of inpainted pairs can improve collaboratively. Overall, the registration of pathological images is achieved in a completely unsupervised learning framework. Experimental results on multiple datasets, including Magnetic Resonance Imaging (MRI) of T1 sequences, demonstrate the efficacy of our proposed method. Our results show that our method can accurately achieve the registration of pathological images and identify lesions even in challenging imaging modalities. Our unsupervised approach offers a promising solution for the efficient and cost-effective registration of pathological images. Our code is available at https://github.com/brain-intelligence-lab/GIRNet.

Irawan C, Iskandar M, Harahap AS et al.

Cosphiadi Irawan,1 Martha Iskandar,1 Agnes Stephanie Harahap,2 Cleopas Martin Rumende,3 Maria Francisca Ham2 1Hematology and Medical Oncology Division, Internal Medicine Department, Dr. Cipto Mangunkusumo Hospital, Faculty of Medicine, Universitas Indonesia, Jakarta, 10430, Indonesia; 2Anatomical Pathology Department, Dr. Cipto Mangunkusumo Hospital, Faculty of Medicine, Universitas Indonesia, Jakarta, 10430, Indonesia; 3Internal Medicine Department, Dr. Cipto Mangunkusumo Hospital, Faculty of Medicine, Universitas Indonesia, Jakarta, 10430, IndonesiaCorrespondence: Martha Iskandar, Tel +628161924095, Email marthaiskandar@gmail.comBackground: Treatment response in diffuse large B-cell lymphoma (DLBCL) is heterogenous. The Hans algorithm (using 30% cut-offs for CD10, BCL6, and MUM1 protein expression) has been the most favored method to categorize DLBCL into germinal center B-cell (GCB) and non-GCB subtypes in order to predict prognosis. However, the algorithm’s ability to prognosticate is not always consistent.Methods: This retrospective cohort study was conducted on DLBCL patients receiving R-CHOP therapy at Dr. Cipto Mangunkusumo Hospital, Jakarta from 2014 to 2017. We aimed to compare the prognostic value of Hans algorithm as well as the protein levels of CD10, BCL6, MUM1, and Ki67 at different cut-offs. Ninety-two patients were classified based on Hans algorithm and various proteins at different cut-off values were analyzed with regard to event-free survival at 24 months using survival analysis. The cut-off values were then compared using receiver operating characteristic curves.Results: A significant survival difference was observed with MUM1 expression cut-off of 50% or more (log rank p = 0.035). CD10, BCL6, Ki67, and Hans algorithm showed AUCs below or near 0.5 (0.405, 0.436, 0.498, and 0.413, respectively), whereas MUM1 showed an AUC of 0.835, in predicting events within 24 months. MUM-1 cut-off of 70.5% yielded an optimal trade-off for sensitivity and specificity.Conclusion: MUM1 expression of 50% or more can help predict prognosis in DLBCL patients receiving R-CHOP therapy and can be considered as for use as a single marker to predict prognosis.Keywords: DLBCL, RCHOP, MUM1, Hans algorithm, prognosisMeSH Terms: lymphoma, large B-cell, dIffuse, MUM1 nucleosome-binding protein, human, antineoplastic combined chemotherapy protocols/adverse effects, rituximab/therapeutic use, retrospective studies, prognosis