Prediction of immunochemotherapy response for diffuse large B-cell lymphoma using artificial intelligence digital pathology.

Diffuse large B-cell lymphoma (DLBCL) is a heterogeneous and prevalent subtype of aggressive non-Hodgkin lymphoma that poses diagnostic and prognostic challenges, particularly in predicting drug responsiveness. In this study, we used digital pathology and deep learning to predict responses to immunochemotherapy in patients with DLBCL. We retrospectively collected 251 slide images from 216 DLBCL patients treated with rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP), with their immunochemotherapy response labels. The digital pathology images were processed using contrastive learning for feature extraction. A multi-modal prediction model was developed by integrating clinical data and pathology image features. Knowledge distillation was employed to mitigate overfitting on gigapixel histopathology images to create a model that predicts responses based solely on pathology images. Based on the importance derived from the attention mechanism of the model, we extracted histological features that were considered key textures associated with drug responsiveness. The multi-modal prediction model achieved an impressive area under the ROC curve of 0.856, demonstrating significant associations with clinical variables such as Ann Arbor stage, International Prognostic Index, and bulky disease. Survival analyses indicated their effectiveness in predicting relapse-free survival. External validation using TCGA datasets supported the model's ability to predict survival differences. Additionally, pathology-based predictions show promise as independent prognostic indicators. Histopathological analysis identified centroblastic and immunoblastic features to be associated with treatment response, aligning with previous morphological classifications and highlighting the objectivity and reproducibility of artificial intelligence-based diagnosis. This study introduces a novel approach that combines digital pathology and clinical data to predict the response to immunochemotherapy in patients with DLBCL. This model shows great promise as a diagnostic and prognostic tool for clinical management of DLBCL. Further research and genomic data integration hold the potential to enhance its impact on clinical practice, ultimately improving patient outcomes.

A comparison between centralized and asynchronous federated learning approaches for survival outcome prediction using clinical and PET data from non-small cell lung cancer patients.

BACKGROUND AND OBJECTIVE: Survival analysis plays an essential role in the medical field for optimal treatment decision-making. Recently, survival analysis based on the deep learning (DL) approach has been proposed and is demonstrating promising results. However, developing an ideal prediction model requires integrating large datasets across multiple institutions, which poses challenges concerning medical data privacy. METHODS: In this paper, we propose FedSurv, an asynchronous federated learning (FL) framework designed to predict survival time using clinical information and positron emission tomography (PET)-based features. This study used two datasets: a public radiogenic dataset of non-small cell lung cancer (NSCLC) from the Cancer Imaging Archive (RNSCLC), and an in-house dataset from the Chonnam National University Hwasun Hospital (CNUHH) in South Korea, consisting of clinical risk factors and F-18 fluorodeoxyglucose (FDG) PET images in NSCLC patients. Initially, each dataset was divided into multiple clients according to histological attributes, and each client was trained using the proposed DL model to predict individual survival time. The FL framework collected weights and parameters from the clients, which were then incorporated into the global model. Finally, the global model aggregated all weights and parameters and redistributed the updated model weights to each client. We evaluated different frameworks including single-client-based approach, centralized learning and FL. RESULTS: We evaluated our method on two independent datasets. First, on the RNSCLC dataset, the mean absolute error (MAE) was 490.80±22.95 d and the C-Index was 0.69±0.01. Second, on the CNUHH dataset, the MAE was 494.25±40.16 d and the C-Index was 0.71±0.01. The FL approach achieved centralized method performance in PET-based survival time prediction and outperformed single-client-based approaches. CONCLUSIONS: Our results demonstrated the feasibility and effectiveness of employing FL for individual survival prediction in NSCLC patients, using clinical information and PET-based features.

Prognostic Significance Of Sequential 18f-fdg Pet/Ct During Frontline Treatment Of Peripheral T Cell Lymphomas.

BACKGROUND/AIMS: The prognostic significance of 18F-fluorodeoxyglucose (FDG)-positron emission tomography-computed tomography (PET/CT) in peripheral T-cell lymphomas (PTCLs) are controversial. We explored the prognostic impact of sequential 18F-FDG PET/CT during frontline chemotherapy of patients with PTCLs. METHODS: In total, 143 patients with newly diagnosed PTCLs were included. Sequential 18F-FDG PET/CTs were performed at the time of diagnosis, during chemotherapy, and at the end of chemotherapy. The baseline total metabolic tumor volume (TMTV) was calculated using the the standard uptake value with a threshold method of 2.5. RESULTS: A baseline TMTV of 457.0 cm3 was used to categorize patients into high and low TMTV groups. Patients with a requirehigh TMTV had shorter progression-free survival (PFS) and overall survival (OS) than those with a low TMTV (PFS, 9.8 vs. 26.5 mo, p = 0.043; OS, 18.9 vs. 71.2 mo, p = 0.004). The interim 18F-FDG PET/CT response score was recorded as 1, 2-3, and 4-5 according to the Deauville criteria. The PFS and OS showed significant differences according to the interim 18F-FDG PET/CT response score (PFS, 120.7 vs. 34.1 vs. 5.1 mo, p < 0.001; OS, not reached vs. 61.1 mo vs. 12.1 mo, p < 0.001). CONCLUSION: The interim PET/CT response based on visual assessment predicts disease progression and survival outcome in PTCLs. A high baseline TMTV is associated with a poor response to anthracycline-based chemotherapy in PTCLs. However, TMTV was not an independent predictor for PFS in the multivariate analysis.

Ensemble Deep Learning Model to Predict Lymphovascular Invasion in Gastric Cancer.

Lymphovascular invasion (LVI) is one of the most important prognostic factors in gastric cancer as it indicates a higher likelihood of lymph node metastasis and poorer overall outcome for the patient. Despite its importance, the detection of LVI(+) in histopathology specimens of gastric cancer can be a challenging task for pathologists as invasion can be subtle and difficult to discern. Herein, we propose a deep learning-based LVI(+) detection method using H&E-stained whole-slide images. The ConViT model showed the best performance in terms of both AUROC and AURPC among the classification models (AUROC: 0.9796; AUPRC: 0.9648). The AUROC and AUPRC of YOLOX computed based on the augmented patch-level confidence score were slightly lower (AUROC: -0.0094; AUPRC: -0.0225) than those of the ConViT classification model. With weighted averaging of the patch-level confidence scores, the ensemble model exhibited the best AUROC, AUPRC, and F1 scores of 0.9880, 0.9769, and 0.9280, respectively. The proposed model is expected to contribute to precision medicine by potentially saving examination-related time and labor and reducing disagreements among pathologists.

Survival Prediction Using Transformer-Based Categorical Feature Representation in the Treatment of Diffuse Large B-Cell Lymphoma.

Diffuse large B-cell lymphoma (DLBCL) is a common and aggressive subtype of lymphoma, and accurate survival prediction is crucial for treatment decisions. This study aims to develop a robust survival prediction strategy to integrate various risk factors effectively, including clinical risk factors and Deauville scores in positron-emission tomography/computed tomography at different treatment stages using a deep-learning-based approach. We conduct a multi-institutional study on 604 DLBCL patients' clinical data and validate the model on 220 patients from an independent institution. We propose a survival prediction model using transformer architecture and a categorical-feature-embedding technique that can handle high-dimensional and categorical data. Comparison with deep-learning survival models such as DeepSurv, CoxTime, and CoxCC based on the concordance index (C-index) and the mean absolute error (MAE) demonstrates that the categorical features obtained using transformers improved the MAE and the C-index. The proposed model outperforms the best-performing existing method by approximately 185 days in terms of the MAE for survival time estimation on the testing set. Using the Deauville score obtained during treatment resulted in a 0.02 improvement in the C-index and a 53.71-day improvement in the MAE, highlighting its prognostic importance. Our deep-learning model could improve survival prediction accuracy and treatment personalization for DLBCL patients.

Deep learning model integrating positron emission tomography and clinical data for prognosis prediction in non-small cell lung cancer patients.

BACKGROUND: Lung cancer is the leading cause of cancer-related deaths worldwide. The majority of lung cancers are non-small cell lung cancer (NSCLC), accounting for approximately 85% of all lung cancer types. The Cox proportional hazards model (CPH), which is the standard method for survival analysis, has several limitations. The purpose of our study was to improve survival prediction in patients with NSCLC by incorporating prognostic information from F-18 fluorodeoxyglucose positron emission tomography (FDG PET) images into a traditional survival prediction model using clinical data. RESULTS: The multimodal deep learning model showed the best performance, with a C-index and mean absolute error of 0.756 and 399 days under a five-fold cross-validation, respectively, followed by ResNet3D for PET (0.749 and 405 days) and CPH for clinical data (0.747 and 583 days). CONCLUSION: The proposed deep learning-based integrative model combining the two modalities improved the survival prediction in patients with NSCLC.

Molecular imaging approaches to facilitate bacteria-mediated cancer therapy.

Bacteria-mediated cancer therapy is a potential therapeutic strategy for cancer that has unique properties, including broad tumor-targeting ability, various administration routes, the flexibility of delivery, and facilitating the host's immune responses. The molecular imaging of bacteria-mediated cancer therapy allows the therapeutically injected bacteria to be visualized and confirms the accurate delivery of the therapeutic bacteria to the target lesion. Several hurdles make bacteria-specific imaging challenging, including the need to discriminate therapeutic bacterial infection from inflammation or other pathologic lesions. To realize the full potential of bacteria-specific imaging, it is necessary to develop bacteria-specific targets that can be associated with an imaging assay. This review describes the current status of bacterial imaging techniques together with the advantages and disadvantages of several imaging modalities. Also, we describe potential targets for bacterial-specific imaging and related applications.

In vivo imaging of invasive aspergillosis with 18F-fluorodeoxysorbitol positron emission tomography.

Invasive aspergillosis is a critical complication in immunocompromised patients with hematologic malignancies or with viral pneumonia caused by influenza virus or SARS­CoV­2. Although early and accurate diagnosis of invasive aspergillosis can maximize clinical outcomes, current diagnostic methods are time-consuming and poorly sensitive. Here, we assess the ability of 2-deoxy-2-18F-fluorosorbitol (18F-FDS) positron emission tomography (PET) to specifically and noninvasively detect Aspergillus infections. We show that 18F-FDS PET can be used to visualize Aspergillus fumigatus infection of the lungs, brain, and muscles in mouse models. In particular, 18F-FDS can distinguish pulmonary aspergillosis from Staphylococcus aureus infection, both of which induce pulmonary infiltrates in immunocompromised patients. Thus, our results indicate that the combination of 18F-FDS PET and appropriate clinical information may be useful in the differential diagnosis and localization of invasive aspergillosis.

Precise characterization of a solitary pulmonary nodule using tumor shadow disappearance rate-corrected F-18 FDG PET and enhanced CT.

ABSTRACT: We aimed to characterize solitary pulmonary nodule (SPN) using imaging parameters for F-18 fluorodeoxyglucose (FDG) positron emission tomography/computed tomography (PET/CT) or enhanced CT corrected by tumor shadow disappearance rate (TDR) to reflect the tissue density.We enrolled 51 patients with an SPN who underwent PET/CT and chest CT with enhancement. The FDG uptake of SPN was evaluated using maximum standardized uptake value (SUVmax) on PET/CT. The mean Hounsfield unit (HU) for each SPN was evaluated over the region of interest on nonenhanced and enhanced CT images. The change in mean HU (HUpeak-pre) was quantified by subtracting the mean HU of the preenhanced CT from that of the post-enhanced CT. TDR was defined as the ratio of the tumor area, which disappears at a mediastinal window, to the tumor area of the lung window. We investigated which parameters (SUVmax or HUpeak-pre) could contribute to the characterization of SPN classified by TDR value and whether diagnostic performance could be improved using TDR-corrected imaging parameters.For SPN with higher tissue density (TDR <42%, n = 22), high value of SUVmax (≥3.1) was a significant factor to predict malignancy (P = .006). High value of HUpeak-pre (≥38) was a significant factor to characterize SPN (P = .002) with lower tissue density (TDR ≥42%, n = 29). The combined approach using TDR-corrected parameters had better predictive performance to characterize SPN than SUVmax only (P = .031).Applying imaging parameters such as SUVmax or HUpeak-pre in consideration of tissue density calculated with TDR could contribute to accurate characterization of SPN.

Bacteria and bacterial derivatives as delivery carriers for immunotherapy.

There is growing interest in the role of microorganisms in human health and disease, with evidence showing that new types of biotherapy using engineered bacterial therapeutics, including bacterial derivatives, can address specific mechanisms of disease. The complex interactions between microorganisms and metabolic/immunologic pathways underlie many diseases with unmet medical needs, suggesting that targeting these interactions may improve patient treatment. Using tools from synthetic biology and chemical engineering, non-pathogenic bacteria or bacterial products can be programmed and designed to sense and respond to environmental signals to deliver therapeutic effectors. This review describes current progress in biotherapy using live bacteria and their derivatives to achieve therapeutic benefits against various diseases.

Effects of Multiple Filters on Liver Tumor Segmentation From CT Images.

Segmentation of liver tumors from Computerized Tomography (CT) images remains a challenge due to the natural variation in tumor shape and structure as well as the noise in CT images. A key assumption is that the performance of liver tumor segmentation depends on the characteristics of multiple features extracted from multiple filters. In this paper, we design an enhanced approach based on a two-class (liver, tumor) convolutional neural network that discriminates tumor as well as liver from CT images. First, the contrast and intensity values in CT images are adjusted and high frequencies are removed using Hounsfield units (HU) filtering and standardization. Then, the liver tumor is segmented from entire images with multiple filter U-net (MFU-net). Finally, a quantitative analysis is carried out to evaluate the segmentation results using three different methods: boundary-distance-based metrics, size-based metrics, and overlap-based metrics. The proposed method is validated on CT images from the 3Dircadb and LiTS dataset. The results demonstrate that the multiple filters are useful for extracting local and global feature simultaneously, minimizing the boundary distance errors, and our approach demonstrates better performance in heterogeneous tumor regions of CT images.

Esophagus Segmentation in CT Images via Spatial Attention Network and STAPLE Algorithm.

One essential step in radiotherapy treatment planning is the organ at risk of segmentation in Computed Tomography (CT). Many recent studies have focused on several organs such as the lung, heart, esophagus, trachea, liver, aorta, kidney, and prostate. However, among the above organs, the esophagus is one of the most difficult organs to segment because of its small size, ambiguous boundary, and very low contrast in CT images. To address these challenges, we propose a fully automated framework for the esophagus segmentation from CT images. The proposed method is based on the processing of slice images from the original three-dimensional (3D) image so that our method does not require large computational resources. We employ the spatial attention mechanism with the atrous spatial pyramid pooling module to locate the esophagus effectively, which enhances the segmentation performance. To optimize our model, we use group normalization because the computation is independent of batch sizes, and its performance is stable. We also used the simultaneous truth and performance level estimation (STAPLE) algorithm to reach robust results for segmentation. Firstly, our model was trained by k-fold cross-validation. And then, the candidate labels generated by each fold were combined by using the STAPLE algorithm. And as a result, Dice and Hausdorff Distance scores have an improvement when applying this algorithm to our segmentation results. Our method was evaluated on SegTHOR and StructSeg 2019 datasets, and the experiment shows that our method outperforms the state-of-the-art methods in esophagus segmentation. Our approach shows a promising result in esophagus segmentation, which is still challenging in medical analyses.

Change of Therapeutic Response Classification According to Recombinant Human Thyrotropin-Stimulated Thyroglobulin Measured at Different Time Points in Papillary Thyroid Carcinoma.

PURPOSE: We investigated whether response classification after total thyroidectomy and radioactive iodine (RAI) therapy could be affected by serum levels of recombinant human thyrotropin (rhTSH)-stimulated thyroglobulin (Tg) measured at different time points in a follow-up of patients with papillary thyroid carcinoma (PTC). METHODS: A total of 147 PTC patients underwent serum Tg measurement for response assessment 6 to 24 months after the first RAI therapy. Serum Tg levels were measured at 24 h (D1Tg) and 48-72 h (D2-3Tg) after the 2nd injection of rhTSH. Responses were classified into three categories based on serum Tg corresponding to the excellent response (ER-Tg), indeterminate response (IR-Tg), and biochemical incomplete response (BIR-Tg). The distribution pattern of response classification based on serum Tg at different time points (D1Tg vs. D2-3Tg) was compared. RESULTS: Serum D2-3Tg level was higher than D1Tg level (0.339 ng/mL vs. 0.239 ng/mL, P < 0.001). The distribution of response categories was not significantly different between D1Tg-based and D2-3Tg-based classification. However, 8 of 103 (7.8%) patients and 3 of 40 (7.5%) patients initially categorized as ER-Tg and IR-Tg based on D1Tg, respectively, were reclassified to IR-Tg and BIR-Tg based on D2-3Tg, respectively. The optimal cutoff values of D1Tg for the change of response categories were 0.557 ng/mL (from ER-Tg to IR-Tg) and 6.845 ng/mL (from IR-Tg to BIR-Tg). CONCLUSION: D1Tg measurement was sufficient to assess the therapeutic response in most patients with low level of D1Tg. Nevertheless, D2-3Tg measurement was still necessary in the patients with D1Tg higher than a certain level as response classification based on D2-3Tg could change.

Correction to: Recent Progress in the Molecular Imaging of Tumor-Treating Bacteria.

[This corrects the article DOI: 10.1007/s13139-021-00689-4.].

Survival time prediction by integrating cox proportional hazards network and distribution function network.

BACKGROUND: The Cox proportional hazards model is commonly used to predict hazard ratio, which is the risk or probability of occurrence of an event of interest. However, the Cox proportional hazard model cannot directly generate an individual survival time. To do this, the survival analysis in the Cox model converts the hazard ratio to survival times through distributions such as the exponential, Weibull, Gompertz or log-normal distributions. In other words, to generate the survival time, the Cox model has to select a specific distribution over time. RESULTS: This study presents a method to predict the survival time by integrating hazard network and a distribution function network. The Cox proportional hazards network is adapted in DeepSurv for the prediction of the hazard ratio and a distribution function network applied to generate the survival time. To evaluate the performance of the proposed method, a new evaluation metric that calculates the intersection over union between the predicted curve and ground truth was proposed. To further understand significant prognostic factors, we use the 1D gradient-weighted class activation mapping method to highlight the network activations as a heat map visualization over an input data. The performance of the proposed method was experimentally verified and the results compared to other existing methods. CONCLUSIONS: Our results confirmed that the combination of the two networks, Cox proportional hazards network and distribution function network, can effectively generate accurate survival time.

Recent Progress in the Molecular Imaging of Tumor-Treating Bacteria.

Bacterial cancer therapy (BCT) approaches have been extensively investigated because bacteria can show unique features of strong tropism for cancer, proliferation inside tumors, and antitumor immunity, while bacteria are also possible agents for drug delivery. Despite the rapidly increasing number of preclinical studies using BCT to overcome the limitations of conventional cancer treatments, very few BCT studies have advanced to clinical trials. In patients undergoing BCT, the precise localization and quantification of bacterial density in different body locations is important; however, most clinical trials have used subjective clinical signs and invasive sampling to confirm bacterial colonization. There is therefore a need to improve the visualization of bacterial densities using noninvasive and repetitive in vivo imaging techniques that can facilitate the clinical translation of BCT. In vivo optical imaging techniques using bioluminescence and fluorescence, which are extensively employed to image the therapeutic process of BCT in small animal research, are hard to apply to the human body because of their low penetrative power. Thus, new imaging techniques need to be developed for clinical trials. In this review, we provide an overview of the various in vivo bacteria-specific imaging techniques available for visualizing tumor-treating bacteria in BCT studies.

Differences in diagnostic impact of dual-tracer PET/computed tomography according to the extrahepatic metastatic site in patients with hepatocellular carcinoma.

OBJECTIVES: We compared the diagnostic performance of C-11 acetate and F-18 fluorodeoxyglucose (FDG) PET/computed tomography (CT) for the detection of extrahepatic metastasis in patients with hepatocellular carcinoma (HCC) and evaluated whether the improvement in the diagnostic performance of dual tracer PET/CT differs by the metastatic site. METHODS: Fifty-eight patients who had extrahepatic metastasis on either C-11 acetate or F-18 FDG PET/CT were enrolled, and 193 metastatic lesions were analyzed in this retrospective study. The metastatic lesions were categorized based on six sites of involvement. According to each involved site, the tracer avidity of the metastatic lesions was compared using the maximum standardized uptake value (SUVmax). RESULTS: Bone was the most frequent categorized metastatic site (44.8%), followed by lymph node (39.7%), lung (34.5%), soft tissue (27.6%), adrenal gland (6.9%), and vascular category (3.4%). C-11 acetate PET/CT showed a higher SUVmax than F-18 FDG PET/CT in metastatic bone lesions (P = 0.003). F-18 FDG uptake was significantly higher than C-11 acetate uptake in metastatic lymph node lesions (P < 0.001). The detection rate of dual tracer PET/CT was significantly higher in the metastatic lung (93.6%) and soft tissue (100%) lesions. However, the diagnostic performance of dual tracer PET/CT was limited in the metastatic bone and lymph node lesions because each tracer's detection rate was very high (bone: 94.6% in C-11 acetate, lymph node: 94.1% in F-18 FDG). CONCLUSIONS: The tracer avidity of metastatic lesions differed according to the involved site. This difference affected the complementary role of dual tracer PET/CT in the diagnosis of extrahepatic metastases in patients with HCC.

Prognostic impact of 18F-FDG PET/CT in patients with multiple myeloma presenting with renal impairment.

Renal insufficiency (RI) is a frequent manifestation of multiple myeloma (MM) at time of diagnosis but there is no reliable prognostic factor for patients with MM presenting with RI. This study investigated the prognostic impact of 18F-fluorodeoxyglucose (18F-FDG) positron emission tomography/computed tomography (PET/CT) in patients with MM with RI at diagnosis. The records of 209 patients with MM between June 2011 and November 2018 were retrospectively analyzed. PET/CT positivity was defined as the presence of more than three focal lesions or the presence of extramedullary disease. Of 209 patients, 90 (43.1%) had RI and showed similar survival outcomes to patients who had normal renal function. In total, 113 patients (54.0%) were PET/CT-positive, and 46.6% of patients with RI were PET/CT-positive at baseline. In patients with RI, those who were PET/CT-positive showed significantly inferior survival outcomes to those who were PET/CT-negative [progression-free survival (PFS), 12.7 vs. 34.0 months, P < 0.001; overall survival (OS), 42.2 months vs. not reached, P = 0.001]. On multivariate analysis, PET/CT positivity was significantly associated with PFS and OS in patients with RI. In conclusion, PET/CT is a reliable imaging technique for predicting survival outcomes in patients with MM with RI.

Clinical Impact of F-18 FDG PET-CT on Biopsy Site Selection in Patients with Suspected Bone Metastasis of Unknown Primary Site.

PURPOSE: We investigated the clinical role of F-18 fluorodeoxyglucose (FDG) positron emission tomography-computed tomography (PET-CT) in the identification of the primary site and the selection of the optimal biopsy site in patients with suspected bone metastasis of unknown primary site. METHODS: The patients with suspected bone metastasis who underwent PET-CT for evaluation of primary site were enrolled in this study. The primary sites were identified by the histopathologic or imaging studies and were classified according to the FDG uptake positivity of the primary site. To evaluate the guiding capability of PET-CT in biopsy site selection, we statistically analyzed whether the biopsy site could be affected according to the presence of extra-skeletal FDG uptake. RESULTS: Among 74 enrolled patients, 51 patients had a metastatic bone disease. The primary site was identified in 48 of 51 patients (94.1%). Forty-six patients were eligible to test the association of clinical choice of biopsy site with PET positivity of extra-skeletal lesion. The extra-skeletal biopsies were done in 42 out of 43 patients with positive extra-skeletal uptake lesions. Bone biopsies were inevitably performed in the other three patients without extra-skeletal uptake lesions. The association came out to be significant (Fisher's exact test, P < 0.001). CONCLUSION: F-18 FDG PET-CT significantly contributed not only to identify the primary site but also to suggest optimal biopsy sites in patients with suspected bone metastasis.