A rat-based preclinical platform facilitating transcatheter hepatic arterial infusion in immunodeficient rats with liver xenografts of patient-derived pancreatic ductal adenocarcinoma.

Liver metastases from pancreatic ductal adenocarcinoma (PDAC) are highly fatal. A rat-based patient-derived tumor xenograft (PDX) model is available for transcatheter therapy. This study aimed to create an immunodeficient rat model with liver xenografts of patient-derived primary PDAC and evaluate efficacy of hepatic arterial infusion chemotherapy with cisplatin in this model. Three patient-derived PDACs were transplanted into the livers of 21 rats each (totally, 63 rats), randomly assigned into hepatic arterial infusion, systemic venous infusion, and control groups (n = 7 each) four weeks post-implantation. Computed tomography evaluated tumor volumes before and four weeks after treatment. Post-euthanasia, resected tumor specimens underwent histopathological examination. A liver-implanted PDAC PDX rat model was established in all 63 rats, with first CT identifying all tumors. Four weeks post-treatment, arterial infusion groups exhibited significantly smaller tumor volumes than controls for all three tumors on second CT. Xenograft tumors histologically maintained adenocarcinoma features compared to original patient tumors. Ki67 expression was significantly lower in arterial infusion groups than in the other two for the three tumors, indicating reduced tumor growth in PDX rats. A liver-implanted PDAC PDX rat model was established as a rat-based preclinical platform. Arterial cisplatin infusion chemotherapy represents a potential therapy for PDAC liver metastasis.

Microinvasion in hepatocellular carcinoma: predictive factor and application for definition of clinical target volume for radiotherapy.

BACKGROUND: To investigate the correlation between microinvasion and various features of hepatocellular carcinoma (HCC), and to clarify the microinvasion distance from visible HCC lesions to subclinical lesions, so as to provide clinical basis for the expandable boundary of clinical target volume (CTV) from gross tumor volume (GTV) in the radiotherapy of HCC. METHODS: HCC patients underwent hepatectomy of liver cancer in our hospital between July 2019 and November 2021 were enrolled. Data on various features and tumor microinvasion distance were collected. The distribution characteristics of microinvasion distance were analyzed to investigate its potential correlation with various features. Tumor size compared between radiographic and pathologic samples was analyzed to clarify the application of pathologic microinvasion to identify subclinical lesions of radiographic imaging. RESULTS: The average microinvasion distance was 0.6 mm, with 95% patients exhibiting microinvasion distance less than 3.0 mm, and the maximum microinvasion distance was 4.0 mm. A significant correlation was found between microinvasion and liver cirrhosis (P = 0.036), serum albumin level (P = 0.049). Multivariate logistic regression analysis revealed that HCC patients with cirrhosis had a significantly lower risk of microinvasion (OR = 0.09, 95%CI = 0.02 ~ 0.50, P = 0.006). Tumor size was overestimated by 1.6 mm (95%CI=-12.8 ~ 16.0 mm) on radiographic size compared to pathologic size, with a mean %Δsize of 2.96% (95%CI=-0.57%~6.50%). The %Δsize ranged from - 29.03% to 34.78%. CONCLUSIONS: CTV expanding by 5.4 mm from radiographic GTV could include all pathologic microinvasive lesions in the radiotherapy of HCC. Liver cirrhosis was correlated with microinvasion and were independent predictive factor of microinvasion in HCC.

Early detection of HCC in patients with cirrhosis using AI; a Systematic Review.

Introduction: Hepatocellular carcinoma constitutes for approximately 75% of primary cancers of liver. Around 80- 90% of patients with HCC have cirrhosis at the time of diagnosis. Use of AI has recently gained significance in the field of hepatology, especially for the detection of HCC, owing to its increasing incidence and specific radiological features which have been established for its diagnostic criteria. Objectives: A systematic review was performed to evaluate the current literature for early diagnosis of hepatocellular carcinoma in cirrhotic patients. METHODS: Systematic review was conducted using PRISMA guidelines and the relevant studies were narrated in detail with assessment of quality for each paper. RESULTS: This systematic review displays the significance of AI in early detection and prognosis of HCC with the pressing need for further exploration in this field. CONCLUSIONS: AI can have a significant role in early diagnosis of HCC in cirrhotic patients.

The Future Role of Abdominal US in Hepatocellular Carcinoma Surveillance.

Abdominal US is currently the best-validated surveillance strategy for hepatocellular carcinoma (HCC) in at-risk patients. It is the only modality shown to have completed all five phases of validation and can achieve high sensitivity and specificity for HCC detection, especially when conducted by expert sonographers in high-volume centers. However, US also has limitations, including operator dependency and varying sensitivity in clinical practice. Further, the sensitivity of US for early-stage HCC detection is lower in patients with obesity or nonviral liver disease, increasingly common populations undergoing surveillance. Imaging-based and blood-based surveillance strategies, including abbreviated MRI and biomarker panels, may overcome some limitations of US-based surveillance. Both strategies have promising test performance in phase II and phase III biomarker studies and are undergoing prospective validation. Considering the variation in HCC risk and test performance between patients, there will likely be a shift away from a one-size-fits-all approach and toward precision screening, in which the "best" test is selected based on individual patient characteristics. In this upcoming era of precision HCC screening among patients with cirrhosis, US will likely continue to have an important, albeit reduced, surveillance role.

Efficacy and Safety of Radiation Segmentectomy with 90Y Resin Microspheres for Hepatocellular Carcinoma.

Background Limited data are available on radiation segmentectomy (RS) for treatment of hepatocellular carcinoma (HCC) using yttrium 90 (90Y) resin microsphere doses determined by using a single-compartment medical internal radiation dosimetry (MIRD) model. Purpose To evaluate the efficacy and safety of RS treatment of HCC with 90Y resin microspheres using a single-compartment MIRD model and correlate posttreatment dose with outcomes. Materials and Methods This retrospective single-center study included adult patients with HCC who underwent RS with 90Y resin microspheres between July 2014 and December 2022. Posttreatment PET/CT and dosimetry were performed. Adverse events were assessed using the Common Terminology Criteria for Adverse Events, version 5.0. Per-lesion and overall response rates (ie, complete response [CR], objective response, disease control, and duration of response) were assessed at imaging using the Modified Response Evaluation Criteria in Solid Tumors, and overall survival (OS) was assessed using Kaplan-Meier analysis. Results Among 67 patients (median age, 69 years [IQR, 63-78 years]; 54 male patients) with HCC, median tumor absorbed dose was 232 Gy (IQR, 163-405 Gy). At 3 months, per-lesion and overall (per-patient) CR was achieved in 47 (70%) and 41 (61%) of 67 patients, respectively. At 6 months (n = 46), per-lesion rates of objective response and disease control were both 94%, and per-patient rates were both 78%. A total of 88% (95% CI: 79 99) and 72% (95% CI: 58, 90) of patients had a per-lesion and overall duration of response of 1 year or greater. At 1 month, a grade 3 clinical adverse event (abdominal pain) occurred in one of 67 (1.5%) patients. Median posttreatment OS was 26 months (95% CI: 20, not reached). Disease progression at 2 years was lower in the group that received 300 Gy or more than in the group that received less than 300 Gy (17% vs 61%; P = .047), with no local progression in the former group through the end of follow-up. Conclusion Among patients with HCC who underwent RS with 90Y resin microspheres, 88% and 72% achieved a per-lesion and overall duration of response of 1 year or greater, respectively, with one grade 3 adverse event. In patients whose tumors received 300 Gy or more according to posttreatment dosimetry, a disease progression benefit was noted. © RSNA, 2024 Supplemental material is available for this article.

Predicting response of hepatoblastoma primary lesions to neoadjuvant chemotherapy through contrast-enhanced computed tomography radiomics.

OBJECTIVE: To investigate the clinical value of contrast-enhanced computed tomography (CECT) radiomics for predicting the response of primary lesions to neoadjuvant chemotherapy in hepatoblastoma. METHODS: Clinical and CECT imaging data were retrospectively collected from 116 children with hepatoblastoma who received neoadjuvant chemotherapy. Tumor response was assessed according to the Response Evaluation Criteria in Solid Tumors (RECIST). Subsequently, they were randomly stratified into a training cohort and a test cohort in a 7:3 ratio. The clinical model was constructed using univariate and multivariate logistic regression, while the radiomics model was developed based on selected radiomics features employing the support vector machine algorithm. The combined clinical-radiomics model incorporated both clinical and radiomics features. RESULTS: The area under the curve (AUC) for the clinical, radiomics, and combined models was 0.704 (95% CI: 0.563-0.845), 0.830 (95% CI: 0.704-0.959), and 0.874 (95% CI: 0.768-0.981) in the training cohort, respectively. In the validation cohort, the combined model achieved the highest mean AUC of 0.830 (95% CI 0.616-0.999), with a sensitivity, specificity, accuracy, precision, and f1 score of 72.0%, 81.1%, 78.5%, 57.2%, and 63.5%, respectively. CONCLUSION: CECT radiomics has the potential to predict primary lesion response to neoadjuvant chemotherapy in hepatoblastoma.

Screening for exclusion of high-risk bleeding features of esophageal varices in cirrhosis through CT and MRI.

BACKGROUND & AIM: Esophageal varices (EV) screening guidelines have evolved with improved risk stratification to avoid unnecessary esophagogastroduodenoscopy (EGD) in individuals with low bleeding risks. However, uncertainties persist in the recommendations for certain patient groups, particularly those with hepatocellular carcinoma (HCC) and/or receiving non-selective beta-blockers (NSBB) without prior endoscopy. This study assessed the efficacy of imaging in ruling out EVs and their high-risk features associated with bleeding in patients with cirrhosis and with HCC. We also evaluated the impact of NSBB on the detection of these characteristics. METHODS: A total of 119 patients undergoing EGD with CT and/or MRI within 90 days of the procedure were included. 87 patients had HCC. A new imaging grading system was developed utilizing the size of EVs and the extent of their protrusion into the esophagus lumen. The negative predictive value (NPV) of EVimaging(-) versus EVimaging (+) (grades 1-3) in ruling out the presence of EV and/or high-risk features by EGD was calculated. The predictive performance of imaging was determined by logistic regression. RESULTS: The NPV of imaging for detecting EV and high-risk features was 81 % and 92 %, respectively. Among HCC patients, the NPV for EV and high-risk features was 80 % and 64 %, respectively. Being on NSBB didn't statistically impact the imaging detection of EV. Imaging was a better predictor of high-risk EGD findings than Child-Turcotte-Pugh scores. CONCLUSIONS: Our results suggest that imaging can effectively rule out the presence of EV and high-risk features during EGD, even in patients with HCC and/or receiving NSBB.

Hepatocellular carcinoma surveillance among people living with hepatitis B in Senegal (SEN-B): insights from a prospective cohort study.

BACKGROUND: Chronic hepatitis B virus (HBV) infection is the predominant cause of hepatocellular carcinoma in west Africa, yet data on the incidence of HBV-related hepatocellular carcinoma remain scarce. We aimed to describe the uptake and early outcomes of systematic ultrasound-based hepatocellular carcinoma screening in SEN-B, which is a prospective HBV cohort in Senegal. METHODS: In this prospective cohort study, we included treatment-naive, HBsAg-positive individuals who were referred to the two infectious diseases clinics (the Department of Tropical and Infectious Diseases and Ambulatory Treatment Center) at Fann University Hospital of Dakar, Senegal, between Oct 1, 2019, and Oct 31, 2022. All participants resided within the Dakar region. Participants underwent abdominal ultrasound, transient elastography, and clinical and virological assessments at inclusion and every 6 months. Liver lesions at least 1 cm in diameter on ultrasound were assessed using four-phase CT, MRI, or liver biopsy. Adherence to hepatocellular carcinoma surveillance was measured using the proportion of time covered, calculated by dividing the cumulative months covered by abdominal ultrasound examinations by the overall follow-up time, defined as the number of months from the date of cohort entry until the last recorded visit, hepatocellular carcinoma diagnosis, or death. Optimal adherence was defined as a proportion of time covered of 100%. FINDINGS: Overall, 755 (99·6%) of 758 participants had at least one abdominal ultrasound performed. The median age of the enrolled participants was 31 years (IQR 25-39), 355 (47·0%) of 755 participants were women, and 82 (10·9%) had a family history of hepatocellular carcinoma. 15 (2·0%) of 755 individuals were HBeAg positive, 206 (27·3%) of 755 individuals had HBV DNA of more than 2000 IU/mL, and 27 (3·6%) of 755 had elastography-defined liver cirrhosis. Of ten (1·3%) participants with a focal lesion at least 1 cm at initial assessment, CT or MRI ruled out hepatocellular carcinoma in nine, whereas imaging and subsequent liver biopsy confirmed one patient with hepatocellular carcinoma. Two further patients with hepatocellular carcinoma were diagnosed at study presentation due to the presence of portal thrombosis on ultrasound. Excluding the three participants with hepatocellular carcinoma identified at baseline, 752 participants were eligible for screening every 6 months. Median follow-up time was 12 months (IQR 6-18) and the median number of ultrasounds per patient was 3 (2-4). During 809·5 person-years of follow-up, one incident hepatocellular carcinoma was reported, resulting in an incidence rate of 1·24 cases per 1000 person-years (95% CI 0·18-8·80). Overall, 702 (93·0%) of 755 participants showed optimal hepatocellular carcinoma surveillance, but this proportion decreased to 77·8% (42 of 54 participants) after 24 months. INTERPRETATION: Hepatocellular carcinoma screening is feasible in HBV research cohorts in west Africa, but its longer-term acceptability needs to be evaluated. Long-term hepatocellular carcinoma incidence data are crucial for shaping tailored screening recommendations. FUNDING: Swiss National Science Foundation, the Swiss Cancer Research Foundation, the National Cancer Institute, and Roche Diagnostics. TRANSLATION: For the French translation of the abstract see Supplementary Materials section.

Controlling Nutritional Status (CONUT) Score is Associated with Overall Survival in Patients with Hepatocellular Carcinoma Treated with Conventional Transcatheter Arterial Chemoembolization: A Propensity Score Matched Analysis.

PURPOSE: This study aims to evaluate the prognostic value of controlling nutritional status (CONUT) score in determining the prognosis of patients with hepatocellular carcinoma (HCC) treated with conventional transcatheter arterial chemoembolization (cTACE). METHODS: This study retrospectively analyzed 936 patients who underwent cTACE for HCC between January 2012 and December 2018, and divided them into two groups based on their CONUT score. To balance the bias in baseline characteristics, propensity score matched (PSM) analysis was conducted. The Kaplan-Meier method was used to establish a cumulative survival curve, and the log-rank test was employed to determine differences in overall survival (OS) and progression-free survival (PFS) among the CONUT score groups. Furthermore, the Cox proportional hazard model was employed to assess the correlation between CONUT score and OS and PFS, whereby hazard ratios (HRs) and 95% confidence intervals (95% CIs) were computed. RESULTS: Before PSM, the median OS for the low (≤ 3) and high (≥ 4) CONUT group (558 vs. 378 patients) was 21.7 and 15.6 months, respectively, and the median PFS was 5.7 and 5 months. Following PSM, both the low and high CONUT score groups comprised 142 patients. The low CONUT score group exhibited a significantly longer OS compared to the high CONUT score group, as determined by the log-rank test (median OS 22.2 vs. 17.0 months, P = 0.014). No significant association was observed between CONUT group and PFS (median PFS 6.4 vs. 4.7 months, log-rank test, P = 0.121). Cox proportional hazard regression analysis revealed that a CONUT score of ≥ 4 was an independent risk factor for OS in patients with HCC who underwent cTACE (HR = 1.361; 95% CI: 1.047-1.771; P = 0.022). These findings were consistent across most subgroup analyses. CONCLUSION: A high CONUT score has been found to be a prognostic factor for poorer OS in patients with HCC who underwent cTACE. LEVEL OF EVIDENCE: Level 3, Non-randomized controlled cohort.

Cascading Detection of Hydrogen Sulfide and N-Acetyltransferase 2 in Hepatocellular Carcinoma Cells Using a Two-Photon Fluorescent Probe.

Hydrogen sulfide (H2S), a critical gas signaling molecule, and N-acetyltransferase 2 (NAT2), a key enzyme in drug metabolism, are both known active biomarkers for liver function. However, the interactions and effects of H2S and NAT2 in living cells or lesion sites remain unknown due to the lack of imaging tools to achieve simultaneous detection of these two substances, making it challenging to implement real-time imaging and precise tracking. Herein, we report an activity-based two-photon fluorescent probe, TPSP-1, for the cascade detection of H2S and NAT2 in living liver cells. Continuous conversion from TPSP-1 to TPSP-3 was achieved in liver cells and tissues. Significantly, leveraging the outstanding optical properties of this two-photon fluorescent probe, TPSP-1, has been effectively used to identify pathological tissue samples directly from clinical liver cancer patients. This work provides us with this novel sensing and two-photon imaging probe, which can be used as a powerful tool to study the physiological functions of H2S and NAT2 and will help facilitate rapid and accurate diagnosis and therapeutic evaluation of hepatocellular carcinoma.

Preclinical evaluation of AGTR1-Targeting molecular probe for colorectal cancer imaging in orthotopic and liver metastasis mouse models.

Despite advancements in colorectal cancer (CRC) treatment, the prognosis remains unfavorable for patients with distant liver metastasis. Fluorescence molecular imaging with specific probes is increasingly used to guide CRC surgical resection in real-time and treatment planning. Here, we demonstrate the targeted imaging capacity of an MPA-PEG4-N3-Ang II probe labeled with near-infrared (NIR) fluorescent dye targeting the angiotensin II (Ang II) type 1 receptor (AGTR1) that is significantly upregulated in CRC. MPA-PEG4-N3-Ang II was highly selective and specific to in vitro tumor cells and in vivo tumors in a mouse CRC xenograft model. The favorable ex vivo imaging and in vivo biodistribution of MPA-PEG4-N3-Ang II afforded tumor-specific accumulation with low background and >10 contrast tumor-to-colorectal values in multiple subcutaneous CRC models at 8 h following injection. Biodistribution analysis confirmed the probe's high uptake in HT29 and HCT116 orthotopic and liver metastatic models of CRC with signal-to-noise ratio (SNR) values of tumor-to-colorectal and -liver fluorescence of 5.8 ± 0.6, 5.3 ± 0.7, and 2.7 ± 0.5, 2.6 ± 0.5, respectively, enabling high-contrast intraoperative tumor visualization for surgical navigation. Given its rapid tumor targeting, precise tumor boundary delineation, durable tumor retention and docking study, MPA-PEG4-N3-Ang II is a promising high-contrast imaging agent for the clinical detection of CRC.

Differences in influencing mechanism of clinicians' adoption behavior for liver cancer screening technology between the leading and subordinate hospitals within medical consortiums.

BACKGROUND: Medical consortiums have been extensively established to facilitate the integration of health resources and bridge the technical gap among member institutions. However, some commonly appropriate technologies remain stagnant in subordinate hospitals, although they have been routinely applied in leading hospitals. Besides, the mechanism underlying differences in clinicians' adoption behavior at different levels of institutions was unknown. Therefore, this study aimed to investigate the differences in influencing mechanisms of clinicians' hepatic contrast-enhanced ultrasound technology (CEUS) utilization behavior between leading and subordinate hospitals within medical consortiums, thus providing clues for expanding effective and appropriate technologies within integrated care systems. METHODS: A self-designed scale was developed based on the theory of planned behavior (TPB). A multistage sampling method was applied to investigate clinicians who were aware of CEUS and worked in liver disease-related departments within the sampled medical institutions. The final sample size was 289. AMOS 24.0 software was used to construct multi-group structural equation modeling (SEM) to validate the hypotheses and determine the mechanism of hepatic CEUS utilization. RESULTS: It revealed that behavioral intention significantly influenced adoption behavior, regardless of whether it was in leading hospitals or subordinate hospitals (ß = 0.283, p < 0.001). Furthermore, behavioral attitude (ß = 0.361, p < 0.001) and perceived behavioral control (ß = 0.582, p < 0.001) exerted significant effects on adoption behavior through behavioral intention. However, in leading hospitals, subjective norm had a significant positive effect on behavioral intention (ß = 0.183, p < 0.01), while it had a significant negative impact on behavioral intention in the subordinate hospitals (ß = -0.348, p < 0.01). CONCLUSION: To effectively translate the adoption intention into actual behavior, it is recommended to elucidate the demand and facilitators involved in the process of health technology adoption across leading and subordinate hospitals. Additionally, bolstering technical support and knowledge dissemination within subordinate hospitals while harnessing the influential role of key individuals can further enhance this transformative process.

Local Tumor Progression Predictive Model Based on MRI for Colorectal Cancer Liver Metastases after Radiofrequency Ablation.

PURPOSE: To investigate the post-radiofrequency ablation (RFA) magnetic resonance imaging (MRI) characteristics in patients with liver metastases from colorectal cancer and to build a predictive model for local tumor progression based on these imaging markers. MATERIALS AND METHODS: A cohort of 73 patients with 110 colorectal cancer liver metastases (CRCLM) who underwent RFA and MRI one month post-ablation was included in image signs analysis and predictive model training. Using a newly developed MRI appearance scoring criteria, MR Image Appearance Scoring at One Month after RFA (MRIAS 1MO), the semi-quantitative analysis of MRI findings within the ablation zone were conducted independently by two radiologists. The intraclass correlation coefficient (ICC) was calculated to evaluate measurement reliability. Differences in MRIAS 1MO scores were compared using Mann-Whitney U test, focusing on local tumor response outcomes. Using local tumor progression (LTP) as the primary end point, MRIAS 1MO scores and other lesion morphological and clinical characteristics were included to establish predictive model. Predication accuracy was subsequently evaluated using calibration curve, time-dependent concordance index (C index) curve, and LTP-free survival (LTPFS) curve. Another cohort comprising 60 patients with 76 CRCLMs provided additional MRIAS 1MO scores and clinical data associated with LTP. We evaluated the performance of the established predictive model using calibration curve, time-dependent C index curve, and LTPFS curve. RESULTS: The MRIAS 1MO criteria exhibited strong measurement reliability. The ICC values of T1S (scores from T1WI), T2S (scores form T2WI) and NCES (scores by adding T1S to T2S) MRIS (the overall scores) were 0.825, 0.779, 0.826 and 0.873, respectively. Lesions with LTP showed significantly higher median values for the overall MRIAS 1MO score (MRIS) compared to lesions without LTP (16 vs. 12, p < 0.001). MRIS and lesion diameter were independent prognostic factors of LTP and were included in predictive model (hazard ratio: MRIS over 13.5:4.275, lesion diameter larger than 30 mm: 2.056). The predictive model demonstrated an overall C index of 0.721 and risk stratification using the predictive model resulted in significantly different LPTFS times. In the validation cohort, the C index were 0.825, 0.794 and 0.764 at six, twelve and twenty-four months, respectively. Patients classified as high-risk in the validation cohort had a median LTPFS time of 10.0 months, while the median LTPFS time was not reached in the low-risk group. CONCLUSIONS: The semi-quantitative MRIAS 1MO criteria, used for post-RFA MRI appearance analysis, exhibited strong measurement reliability. Prediction models established based on overall MRIAS 1MO score (MRIS) and lesion diameter had good predictive performance for LTP in patients undergoing RFA for CRCLM treatment.

S2DA-Net: Spatial and spectral-learning double-branch aggregation network for liver tumor segmentation in CT images.

Accurate liver tumor segmentation is crucial for aiding radiologists in hepatocellular carcinoma evaluation and surgical planning. While convolutional neural networks (CNNs) have been successful in medical image segmentation, they face challenges in capturing long-term dependencies among pixels. On the other hand, Transformer-based models demand a high number of parameters and involve significant computational costs. To address these issues, we propose the Spatial and Spectral-learning Double-branched Aggregation Network (S2DA-Net) for liver tumor segmentation. S2DA-Net consists of a double-branched encoder and a decoder with a Group Multi-Head Cross-Attention Aggregation (GMCA) module, Two branches in the encoder consist of a Fourier Spectral-learning Multi-scale Fusion (FSMF) branch and a Multi-axis Aggregation Hadamard Attention (MAHA) branch. The FSMF branch employs a Fourier-based network to learn amplitude and phase information, capturing richer features and detailed information without introducing an excessive number of parameters. The FSMF branch utilizes a Fourier-based network to capture amplitude and phase information, enriching features without introducing excessive parameters. The MAHA branch incorporates spatial information, enhancing discriminative features while minimizing computational costs. In the decoding path, a GMCA module extracts local information and establishes long-term dependencies, improving localization capabilities by amalgamating features from diverse branches. Experimental results on the public LiTS2017 liver tumor datasets show that the proposed segmentation model achieves significant improvements compared to the state-of-the-art methods, obtaining dice per case (DPC) 69.4 % and global dice (DG) 80.0 % for liver tumor segmentation on the LiTS2017 dataset. Meanwhile, the pre-trained model based on the LiTS2017 datasets obtain, DPC 73.4 % and an DG 82.2 % on the 3DIRCADb dataset.

Tumor conspicuity enhancement-based segmentation model for liver tumor segmentation and RECIST diameter measurement in non-contrast CT images.

BACKGROUND AND OBJECTIVE: Liver tumor segmentation (LiTS) accuracy on contrast-enhanced computed tomography (CECT) images is higher than that on non-contrast computed tomography (NCCT) images. However, CECT requires contrast medium and repeated scans to obtain multiphase enhanced CT images, which is time-consuming and cost-increasing. Therefore, despite the lower accuracy of LiTS on NCCT images, which still plays an irreplaceable role in some clinical settings, such as guided brachytherapy, ablation, or evaluation of patients with renal function damage. In this study, we intend to generate enhanced high-contrast pseudo-color CT (PCCT) images to improve the accuracy of LiTS and RECIST diameter measurement on NCCT images. METHODS: To generate high-contrast CT liver tumor region images, an intensity-based tumor conspicuity enhancement (ITCE) model was first developed. In the ITCE model, a pseudo color conversion function from an intensity distribution of the tumor was established, and it was applied in NCCT to generate enhanced PCCT images. Additionally, we design a tumor conspicuity enhancement-based liver tumor segmentation (TCELiTS) model, which was applied to improve the segmentation of liver tumors on NCCT images. The TCELiTS model consists of three components: an image enhancement module based on the ITCE model, a segmentation module based on a deep convolutional neural network, and an attention loss module based on restricted activation. Segmentation performance was analyzed using the Dice similarity coefficient (DSC), sensitivity, specificity, and RECIST diameter error. RESULTS: To develop the deep learning model, 100 patients with histopathologically confirmed liver tumors (hepatocellular carcinoma, 64 patients; hepatic hemangioma, 36 patients) were randomly divided into a training set (75 patients) and an independent test set (25 patients). Compared with existing tumor automatic segmentation networks trained on CECT images (U-Net, nnU-Net, DeepLab-V3, Modified U-Net), the DSCs achieved on the enhanced PCCT images are both improved compared with those on NCCT images. We observe improvements of 0.696-0.713, 0.715 to 0.776, 0.748 to 0.788, and 0.733 to 0.799 in U-Net, nnU-Net, DeepLab-V3, and Modified U-Net, respectively, in terms of DSC values. In addition, an observer study including 5 doctors was conducted to compare the segmentation performance of enhanced PCCT images with that of NCCT images and showed that enhanced PCCT images are more advantageous for doctors to segment tumor regions. The results showed an accuracy improvement of approximately 3%-6%, but the time required to segment a single CT image was reduced by approximately 50 %. CONCLUSIONS: Experimental results show that the ITCE model can generate high-contrast enhanced PCCT images, especially in liver regions, and the TCELiTS model can improve LiTS accuracy in NCCT images.

A multi-instance tumor subtype classification method for small PET datasets using RA-DL attention module guided deep feature extraction with radiomics features.

BACKGROUND: Positron emission tomography (PET) is extensively employed for diagnosing and staging various tumors, including liver cancer, lung cancer, and lymphoma. Accurate subtype classification of tumors plays a crucial role in formulating effective treatment plans for patients. Notably, lymphoma comprises subtypes like diffuse large B-cell lymphoma and Hodgkin's lymphoma, while lung cancer encompasses adenocarcinoma, small cell carcinoma, and squamous cell carcinoma. Similarly, liver cancer consists of subtypes such as cholangiocarcinoma and hepatocellular carcinoma. Consequently, the subtype classification of tumors based on PET images holds immense clinical significance. However, in clinical practice, the number of cases available for each subtype is often limited and imbalanced. Therefore, the primary challenge lies in achieving precise subtype classification using a small dataset. METHOD: This paper presents a novel approach for tumor subtype classification in small datasets using RA-DL (Radiomics-DeepLearning) attention. To address the limited sample size, Support Vector Machines (SVM) is employed as the classifier for tumor subtypes instead of deep learning methods. Emphasizing the importance of texture information in tumor subtype recognition, radiomics features are extracted from the tumor regions during the feature extraction stage. These features are compressed using an autoencoder to reduce redundancy. In addition to radiomics features, deep features are also extracted from the tumors to leverage the feature extraction capabilities of deep learning. In contrast to existing methods, our proposed approach utilizes the RA-DL-Attention mechanism to guide the deep network in extracting complementary deep features that enhance the expressive capacity of the final features while minimizing redundancy. To address the challenges of limited and imbalanced data, our method avoids using classification labels during deep feature extraction and instead incorporates 2D Region of Interest (ROI) segmentation and image reconstruction as auxiliary tasks. Subsequently, all lesion features of a single patient are aggregated into a feature vector using a multi-instance aggregation layer. RESULT: Validation experiments were conducted on three PET datasets, specifically the liver cancer dataset, lung cancer dataset, and lymphoma dataset. In the context of lung cancer, our proposed method achieved impressive performance with Area Under Curve (AUC) values of 0.82, 0.84, and 0.83 for the three-classification task. For the binary classification task of lymphoma, our method demonstrated notable results with AUC values of 0.95 and 0.75. Moreover, in the binary classification task of liver tumor, our method exhibited promising performance with AUC values of 0.84 and 0.86. CONCLUSION: The experimental results clearly indicate that our proposed method outperforms alternative approaches significantly. Through the extraction of complementary radiomics features and deep features, our method achieves a substantial improvement in tumor subtype classification performance using small PET datasets.

Markerless liver online adaptive stereotactic radiotherapy: feasibility analysisCervantes.

Objective. Radio-opaque markers are recommended for image-guided radiotherapy in liver stereotactic ablative radiotherapy (SABR), but their implantation is invasive. We evaluate in thisin-silicostudy the feasibility of cone-beam computed tomography-guided stereotactic online-adaptive radiotherapy (CBCT-STAR) to propagate the target volumes without implanting radio-opaque markers and assess its consequence on the margin that should be used in that context.Approach. An emulator of a CBCT-STAR-dedicated treatment planning system was used to generate plans for 32 liver SABR patients. Three target volume propagation strategies were compared, analysing the volume difference between the GTVPropagatedand the GTVConventional, the vector lengths between their centres of mass (lCoM), and the 95th percentile of the Hausdorff distance between these two volumes (HD95). These propagation strategies were: (1) structure-guided deformable registration with deformable GTV propagation; (2) rigid registration with rigid GTV propagation; and (3) image-guided deformable registration with rigid GTV propagation. Adaptive margin calculation integrated propagation errors, while interfraction position errors were removed. Scheduled plans (PlanNon-adaptive) and daily-adapted plans (PlanAdaptive) were compared for each treatment fraction.Main results.The image-guided deformable registration with rigid GTV propagation was the best propagation strategy regarding tolCoM(mean: 4.3 +/- 2.1 mm), HD95 (mean 4.8 +/- 3.2 mm) and volume preservation between GTVPropagatedand GTVConventional. This resulted in a planning target volume (PTV) margin increase (+69.1% in volume on average). Online adaptation (PlanAdaptive) reduced the violation rate of the most important dose constraints ('priority 1 constraints', 4.2 versus 0.9%, respectively;p< 0.001) and even improved target volume coverage compared to non-adaptive plans (PlanNon-adaptive).Significance. Markerless CBCT-STAR for liver tumours is feasible using Image-guided deformable registration with rigid GTV propagation. Despite the cost in terms of PTV volumes, daily adaptation reduces constraints violation and restores target volumes coverage.