Diagnosis of liver tumors by multimodal ultrasound imaging

Towards robust multimodal ultrasound classification for liver tumor diagnosis: A generative approach to modality missingness

BACKGROUND AND OBJECTIVE

In medical image analysis, combining multiple imaging modalities enhances diagnostic accuracy by providing complementary information. However, missing modalities are common in clinical settings, limiting the effectiveness of multimodal models. This study addresses the challenge of missing modalities in liver tumor diagnosis by proposing a generative model-based method for cross-modality reconstruction and classification. The dataset for this study comprises 359 case data from a hospital, with each case including three modality data: B-mode ultrasound images, Color Doppler Flow Imaging (CDFI), and clinical data. Only cases with one missing image modality are considered, excluding those with missing clinical data.

METHODS

We developed a multimodal classification framework specifically for liver tumor diagnosis, employing various feature extraction networks to explore the impact of different modality combinations on classification performance when only available modalities are used. DenseNet extracts CDFI features, while EfficientNet is employed for B-mode ultrasound image feature extraction. These features are then flattened and concatenated with clinical data using feature-level fusion to obtain a full-modality model. Modality weight parameters are introduced to emphasize the importance of different modalities, yielding Model_D, which serves as the classification model after subsequent image modality supplementation. In cases of missing modalities, generative models, including U-GAT-IT and MSA-GAN, are utilized for cross-modal reconstruction of missing B-mode ultrasound or CDFI images (e.g., reconstructing CDFI from B-mode ultrasound when CDFI is missing). After evaluating the usability of the generated images, they are input into Model_D as supplementary images for the missing modalities.

RESULTS

Model performance and modality supplementation effects were evaluated through accuracy, precision, recall, F1 score, and AUC metrics. The results demonstrate that the proposed Model_D, which introduces modality weights, achieves an accuracy of 88.57 %, precision of 87.97 %, recall of 82.32 %, F1 score of 0.87, and AUC of 0.95 in the full-modality classification task for liver tumors. Moreover, images reconstructed using U-GAT-IT and MSA-GAN across modalities exhibit PSNR > 20 and multi-scale structural similarity > 0.7, indicating moderate image quality with well-preserved overall structures, suitable for input into the model as supplementary images in cases of missing modalities. The supplementary CDFI or B-mode ultrasound images achieve 87.10 % and 86.43 % accuracy, respectively, with AUC values of 0.92 and 0.95. This proves that even in the absence of certain modalities, the generative models can effectively reconstruct missing images, maintaining high classification performance comparable to that in complete modality scenarios.

CONCLUSIONS

The generative model-based approach for modality reconstruction significantly improves the robustness of multimodal classification models, particularly in the context of liver tumor diagnosis. This method enhances the clinical applicability of multimodal models by ensuring high diagnostic accuracy despite missing modalities. Future work will explore further improvements in modality reconstruction techniques to increase the generalization and reliability of the model in various clinical settings.

Using multimodal ultrasound technology to improve the success rate of liver tumor puncture for lesions with poor visibility on conventional ultrasound imaging

Background

The ability of conventional ultrasound (US)-guided liver biopsy to visualize certain liver lesions, particularly those affected by conditions like hepatitis or cirrhosis, which can obscure lesion boundaries and lead to inaccurate biopsy targeting, is limited. This study aimed to evaluate the potential of multimodal US techniques to improve the visibility of liver lesions that are indistinct under conventional US, and to enhance the success rate of percutaneous biopsies.

Methods

In total, 144 patients with liver masses and lesions that were not clearly visible on conventional US from October 2018 to January 2024 were enrolled in this retrospective analysis. The lesions of these patients exhibited poor visibility on conventional US, but the tumor location was visible on abdominal computerized tomography (CT) or magnetic resonance (MR) imaging scans. After excluding patients who did not undergo biopsy or patients with lesions that were remained not clearly visible on multimodal US examinations. Ultimately, a total of 95 patients were enrolled in this study. We analyzed the clinical and imaging data for all these patients. CT/MR-US fusion imaging was performed in 55 patients, contrast-enhanced ultrasound (CEUS) was performed in 95 patients, and high-frequency US was performed in 21 patients. The visibility of the lesions using these three techniques was evaluated, and the consistency between the biopsy pathology and the final diagnosis was analyzed.

Results

In the study, the detection rates of lesions using CT/MR-US fusion imaging, CEUS, and high-frequency US were 49.1%, 96.8%, and 76.2%, respectively. After confirming the target location of the lesions, all patients underwent percutaneous US-guided biopsy. The accuracy rate of the biopsies was 91.6%, and the positive concordance rate was 91.1%. Among the 13 patients with negative pathology findings after biopsy, 8 had false-negative results (based on follow-up laboratory tests and imaging results consistent with malignant tumor characteristics), resulting in a false-negative rate of 8.9%.

Conclusions

Multimodal US significantly improved the success of biopsies for liver lesions not clearly visible on conventional US, aiding in precise treatment planning.

Multimodal ultrasound plus tumor markers demonstrates a high value in enhanced diagnosis of breast cancer

OBJECTIVE

To determine the diagnostic value of multimodal ultrasound combined with tumor markers in breast cancer (BC).

METHODS

A retrospective analysis was conducted on 198 patients with breast lesions treated at the Affiliated Wuxi People's Hospital of Nanjing Medical University between May 2020 and May 2023. All patients underwent multimodal ultrasound and tumor marker testing. Among the 198 patients, 88 patients were pathologically diagnosed with benign disease (benign group) and 110 patients were pathologically diagnosed with malignant disease (malignant group). With the pathological results as the gold standard, the benign and malignant results from different diagnostic methods were compared, focusing on specificity, sensitivity and accuracy.

RESULTS

The areas under the curves (AUCs) of carbohydrate antigen 153 (CA153), CA125, and carcinoembryonic antigen (CEA) for diagnosing BC were 0.810, 0.812, and 0.790, respectively. When these tumor markers were used in combination for diagnosing BC, the AUC increased to 0.928. The AUC of multimodal ultrasound alone in diagnosing BC was 0.845. Additionally, the AUC of multimodal ultrasound combined with tumor markers in diagnosing BC reached 0.971, with the corresponding specificity, sensitivity and accuracy of 90.00%, 94.43% and 91.92%, respectively.

CONCLUSION

In patients with early BC, the combination of multimodal ultrasound and tumor marker detection significantly improves the accuracy of diagnosing benign and malignant breast lesions compared to using either modality alone.

Diagnostic value of multimodal ultrasound for breast cancer and prediction of sentinel lymph node metastases

Background

Sentinel lymph node metastasis (SLNM) is a critical factor in the prognosis and treatment planning for breast cancer (BC), as it indicates the potential spread of cancer to other parts of the body. The accurate prediction and diagnosis of SLNM are essential for improving clinical outcomes and guiding treatment decisions.

Objective

This study aimed to construct a Lasso regression model by integrating multimodal ultrasound (US) techniques, including US, shear wave elastography (SWE), and contrast-enhanced ultrasound (CEUS), to improve the predictive accuracy of sentinel lymph node metastasis in breast cancer and provide more precise guidance for clinical treatment.

Results

A total of 253 eligible samples were screened, of which 148 were group benign and 105 were group malignant. There were statistically significant differences (p < 0.05) between group malignant patients in terms of age, palpable mass, body mass index, distance to nipple, maximum diameter, blood flow, microcalcification, 2D border, 2D morphology, and 2D uniformity and group benign. The Lasso regression model was useful in the diagnosis of benign and malignant nodules with an AUC of 0.966 and in diagnosing SLNM with an AUC of 0.832.

Conclusion

In this study, we successfully constructed and validated a Lasso regression model based on the multimodal ultrasound technique for predicting whether SLNM occurs in BCs, showing high diagnostic accuracy.

Diagnostic Performance of an Artificial Intelligence Model Based on Contrast-Enhanced Ultrasound in Patients with Liver Lesions: A Comparative Study with Clinicians

Contrast-enhanced ultrasound (CEUS) is widely used in the characterization of liver tumors; however, the evaluation of perfusion patterns using CEUS has a subjective character. This study aims to evaluate the accuracy of an automated method based on CEUS for classifying liver lesions and to compare its performance with that of two experienced clinicians. The system used for automatic classification is based on artificial intelligence (AI) algorithms. For an interpretation close to the clinical setting, both clinicians knew which patients were at high risk for hepatocellular carcinoma (HCC), but only one was aware of all the clinical data. In total, 49 patients with 59 liver tumors were included. For the benign and malignant classification, the AI model outperformed both clinicians in terms of specificity (100% vs. 93.33%); still, the sensitivity was lower (74% vs. 93.18% vs. 90.91%). In the second stage of multiclass diagnosis, the automatic model achieved a diagnostic accuracy of 69.93% for HCC and 89.15% for liver metastases. Readers demonstrated greater diagnostic accuracy for HCC (83.05% and 79.66%) and liver metastases (94.92% and 96.61%) compared to the AI system; however, both were experienced sonographers. The AI model could potentially assist and guide less-experienced clinicians to discriminate malignant from benign liver tumors with high accuracy and specificity.

Shear-wave elastography combined with contrast-enhanced ultrasound algorithm for noninvasive characterization of focal liver lesions

PURPOSE

To establish shear-wave elastography (SWE) combined with contrast-enhanced ultrasound (CEUS) algorithm (SCCA) and improve the diagnostic performance in differentiating focal liver lesions (FLLs).

MATERIAL AND METHODS

We retrospectively selected patients with FLLs between January 2018 and December 2019 at the First Affiliated Hospital of Sun Yat-sen University. Histopathology was used as a standard criterion except for hemangiomas and focal nodular hyperplasia. CEUS with SonoVue (Bracco Imaging) and SCCA combining CEUS and maximum value of elastography with < 20 kPa and > 90 kPa thresholds were used for the diagnosis of FLLs. The diagnostic performance of CEUS and SCCA was calculated and compared.

RESULTS

A total of 171 FLLs were included, with 124 malignant FLLs and 47 benign FLLs. The area under curve (AUC), sensitivity, and specificity in detecting malignant FLLs were 0.83, 91.94%, and 74.47% for CEUS, respectively, and 0.89, 91.94%, and 85.11% for SCCA, respectively. The AUC of SCCA was significantly higher than that of CEUS (P = 0.019). Decision curves indicated that SCCA provided greater clinical benefits. The SCCA provided significantly improved prediction of clinical outcomes, with a net reclassification improvement index of 10.64% (P = 0.018) and integrated discrimination improvement of 0.106 (P = 0.019). For subgroup analysis, we divided the FLLs into a chronic-liver-disease group (n = 88 FLLs) and a normal-liver group (n = 83 FLLs) according to the liver background. In the chronic-liver-disease group, there were no differences between the CEUS-based and SCCA diagnoses. In the normal-liver group, the AUC of SCCA and CEUS in the characterization of FLLs were 0.89 and 0.83, respectively (P = 0.018).

CONCLUSION

SCCA is a feasible tool for differentiating FLLs in patients with normal liver backgrounds. Further investigations are necessary to validate the universality of this algorithm.

Deep Learning Algorithms in the Automatic Segmentation of Liver Lesions in Ultrasound Investigations

BACKGROUND

The ultrasound is one of the most used medical imaging investigations worldwide. It is non-invasive and effective in assessing liver tumors or other types of parenchymal changes.

METHODS

The aim of the study was to build a deep learning model for image segmentation in ultrasound video investigations. The dataset used in the study was provided by the University of Medicine and Pharmacy Craiova, Romania and contained 50 video examinations from 49 patients. The mean age of the patients in the cohort was 69.57. Regarding presence of a subjacent liver disease, 36.73% had liver cirrhosis and 16.32% had chronic viral hepatitis (5 patients: chronic hepatitis C and 3 patients: chronic hepatitis B). Frames were extracted and cropped from each examination and an expert gastroenterologist labelled the lesions in each frame. After labelling, the labels were exported as binary images. A deep learning segmentation model (U-Net) was trained with focal Tversky loss as a loss function. Two models were obtained with two different sets of parameters for the loss function. The performance metrics observed were intersection over union and recall and precision.

RESULTS

Analyzing the intersection over union metric, the first segmentation model obtained performed better compared to the second model: 0.8392 (model 1) vs. 0.7990 (model 2). The inference time for both models was between 32.15 milliseconds and 77.59 milliseconds.

CONCLUSIONS

Two segmentation models were obtained in the study. The models performed similarly during training and validation. However, one model was trained to focus on hard-to-predict labels. The proposed segmentation models can represent a first step in automatically extracting time-intensity curves from CEUS examinations.

Multimodal Imaging under Artificial Intelligence Algorithm for the Diagnosis of Liver Cancer and Its Relationship with Expressions of EZH2 and p57

Objective

It aimed to explore the diagnostic efficacy of multimodal ultrasound images based on mask region with convolutional neural network (M-RCNN) segmentation algorithm for small liver cancer and analyze the expression of zeste gene enhancer homolog 2 (EZH2) and p57 (P57 Kip2) genes in cancer cells.

Methods

A total of 100 patients suspected of small liver cancer were randomly divided into Doppler group (color Doppler ultrasound examination), contrast group (contrast ultrasound examination), elastic group (ultrasound elastography examination), and multimodal group (combined examination of the three methods), with 25 patients in each group. Images were processed by the M-RCNN segmentation algorithm. The results of the pathological biopsy were used to evaluate the diagnostic efficacy of the four methods. The liver tissues were then extracted and divided into observation group 1 (lesion tissue specimen), observation group 2 (liver tissue around cancer lesion), and control group (normal liver tissue), and the expression activities of EZH2 and p57 genes in the three groups were analyzed.

Results

The accuracy of M-RCNN (97.23%) and average precision (AP) (71.90%) were higher than other methods (P < 0.05). Sensitivity (88.87%), specific degree of consistency (90.91%), accuracy (89.47%), and consistence (0.68) of the multimodal group were better than the other three groups (P < 0.05). Low and medium differentiated cancer tissues had an irregular shape, unclear boundary, uneven internal echo, unchanged/enhanced posterior echo, blood flow level 1∼2, elastic score 4∼5, and enhancement mode fast in and fast out. The positive expression rate of EZH2 in observation group 1 (75.95%) was higher than that in the other two groups, the positive expression rate of p57 in observation group 1 (80.79%) was lower than that in the other two groups, and the positive expression rate of p57 in the highly differentiated cancer foci (80.79%) was significantly lower than that in the middle and low differentiated cancer foci (P < 0.05).

Conclusions

M-RCNN segmentation algorithm had a better segmentation effect. Multimodal ultrasound had a good effect on the benign and malignant diagnosis of small liver cancer and had a high clinical application value. The high expression of EZH2 and the decreased expression of p57 can promote the occurrence of small hepatocellular carcinoma, and the deficiency of the P57 gene was related to the low differentiation of cancer cells.

Application of Shear Wave Elastography and Contrast-Enhanced Ultrasound in Transrectal Prostate Biopsy

OBJECTIVE

To explore the clinical value of ultrasound shear wave elastography (SWE) and contrast-enhanced ultrasound (CEUS) in transrectal prostate biopsy.

METHODS

A total of 54 patients (average age: 67.79±12.01 years) in the experimental group underwent transrectal prostate biopsy under the guidance of SWE, while 46 patients (average age: 69.22±11.54 years) in the control group underwent transrectal prostate biopsy guided by CEUS.

RESULTS

There were a total of 451 needles, with an average of 8.35±1.67 needles per patient in the experimental group, and a total of 462 needles, with an average of 10.04±1.33 needles per patient in the control group. The difference in puncture times between the two groups was statistically significant (P<0.05). There was no significant difference in the positive detection rate, sensitivity or specificity between the two groups (P>0.05), but there was a significant difference in the diagnostic accuracy between the two groups (P<0.05). The E_mean and E_max of prostate cancer were significantly higher in the experimental group than in benign prostatic hyperplasia (P<0.05). The receiver operating characteristic curve (ROC) analysis showed that the area under the ROC curve (AUC) of E_mean was 0.752 (S.E. =0.072, 95% CI=0.611-0.894, P=0.007), and the best cutoff value was 47.005 kPa.

CONCLUSION

In summary, both SWE- and CEUS-guided transrectal prostate biopsy can help find the focus and guide the puncture, and improve the positive detection rate.

The value of multimodal ultrasonography in differential diagnosis of tuberculous and non-tuberculous superficial lymphadenitis

Background

To investigate the value of multimodal ultrasonography in differentiating tuberculosis from other lymphadenopathy.

Methods

Sixty consecutive patients with superficial lymphadenopathy treated at our hospital from January 2017 to December 2018 were categorized into four types based on the color Doppler ultrasound, five types based on contrast-enhanced ultrasound, and five types based on elastography. Sensitivity and specificity were calculated of all the three imaging, including color Doppler examination, contrast-enhanced ultrasound and one individual multimodal method, for detecting lymph nodes.

Results

A total of 60 patients were included in the final analysis. Of those, Mycobacterium tuberculosis was positive in 38 patients and negative in 22 patients. Among the 38 patients who were positive for Mycobacterium tuberculosis, of which 23 had a history of pulmonary tuberculosis, accounting for 60.53% of the positive cases, and the remaining patients did not combine lesions of other organs. Among the 60 superficial lymph nodes, 63.3% presented with tuberculous lymphadenitis. The sensitivity, specificity, and accuracy of the color Doppler examination were 73.68%, 68.18%, and 71.67%, respectively. The sensitivity, specificity and accuracy of contrast-enhanced ultrasound were 89.47%, 63.64% and 80.00%, respectively. The sensitivity, specificity and accuracy of the elastography were 63.16%, 63.64% and 63.33%, respectively. The sensitivity, specificity and accuracy of one individual multimodal method were 42.11%, 95.45% and 61.67%, respectively. The sensitivity, specificity and accuracy of all modes combined were 100.00%, 27.27% and 73.33%, respectively.

Conclusion

Multimodal ultrasonography has high predictive value for the differential diagnosis of superficial tuberculous lymphadenitis.

Molecular Photoacoustic Imaging

Medicine has gone through several challenges to make it much more accurate and thus prolong the human being's life. A large part of this challenge is diseased, so early detection can help carry out treatment on time. There is a technology that allows detecting an abnormality within the body without using an invasive method. Ultrasound is a diagnostic test used to scan organs and tissues through sound waves. Although this technique has been widely used, the results are not desired because the images generated are not high resolution. On the other hand, X-rays are used because it presents an image with a much higher resolution than other techniques based on light waves or ultrasound; despite this, they are harmful to cells. In consequence of this problem, another method called molecular photoacoustic imaging has been implemented. This technique bridges the traditional depth limits of ballistic optical imaging and diffuse optical imaging's resolution limits, using the acoustic waves generated in response to laser light absorption, which has now shown potential for molecular imaging, allowing the visualization of biological processes in a non-invasive way. The purpose of this article is to give a critically scoped review of the physical, chemical, and biochemical characteristics of existing photoacoustic contrast agents, highlighting the pivotal applications and current challenges for molecular photoacoustic imaging.

The Need for Alternatives to Liver Biopsies: Non-Invasive Analytics and Diagnostics

Histology remains essential for the diagnosis and management of many disorders affecting the liver. However, the biopsy procedure itself is associated with a low risk of harm to the patient and cost to the health services; samples may not be adequate and are subject to sampling variation. Furthermore, interpretation often depends on the skill of the pathologist. Increasingly, new techniques are becoming available that are altering the indications for liver biopsy. Many diseases of the liver can be diagnosed and managed using serological and radiological techniques; the degree of fibrosis and fat can often be assessed by serological or imaging techniques and the nature of space occupying lesions defined by serology, imaging and use of liquid biopsy. However, these techniques, too, are subject to limitations: sensitivity and specificity is not always adequate for diagnosis or management; some techniques are expensive and often also require expert interpretation. Although there may be less need for liver biopsy today, histology remains the gold standard as well as an essential tool for the diagnosis and management of many conditions, especially where there are multiple pathologies, or where a diagnosis cannot or has not been made by alternative approaches. Until less invasive techniques become more reliable and accessible, liver histology will remain a key investigation.

An Improved Sensing Method of a Robotic Ultrasound System for Real-Time Force and Angle Calibration

An ultrasonic examination is a clinically universal and safe examination method, and with the development of telemedicine and precision medicine, the robotic ultrasound system (RUS) integrated with a robotic arm and ultrasound imaging system receives increasing attention. As the RUS requires precision and reproducibility, it is important to monitor the real-time calibration of the RUS during examination, especially the angle of the probe for image detection and its force on the surface. Additionally, to speed up the integration of the RUS and the current medical ultrasound system (US), the current RUSs mostly use a self-designed fixture to connect the probe to the arm. If the fixture has inconsistencies, it may cause an operating error. In order to improve its resilience, this study proposed an improved sensing method for real-time force and angle calibration. Based on multichannel pressure sensors, an inertial measurement unit (IMU), and a novel sensing structure, the ultrasonic probe and robotic arm could be simply and rapidly combined, which rendered real-time force and angle calibration at a low cost. The experimental results show that the average success rate of the downforce position identification achieved was 88.2%. The phantom experiment indicated that the method could assist the RUS in the real-time calibration of both force and angle during an examination.

Knockdown of lncRNA LINC01234 Suppresses the Tumorigenesis of Liver Cancer via Sponging miR-513a-5p

Background

Liver cancer is a frequent malignancy with poor prognosis and high mortality all over the world. It has been reported many lncRNAs could modulate the tumorigenesis of liver cancer. To identify novel potential targets for liver cancer, the differential expressed lncRNAs between liver cancer and adjacent normal tissues was analyzed with bioinformatics tool.

Methods

The differential expressed lncRNAs between liver cancer and adjacent normal tissues were analyzed with bioinformatics tool. Cell viability and proliferation was tested by CCK8 and Ki67, respectively. Apoptosis of liver cancer cells was tested by flow cytometry. Gene and protein expressions in liver cancer cells were measured by qRT-PCR and western blot, respectively. In vivo model of liver cancer was established to detect the effect of LINC01234 on liver cancer in vivo.

Results

LINC01234 was found to be negatively correlated with the survival rate of patients with liver cancer. Moreover, knockdown of LINC01234 significantly suppressed the proliferation and invasion of liver cancer cells via inducing the apoptosis. Meanwhile, miR-513a-5p was sponged by LINC01234, and USP4 was found to be a direct target of miR-513a-5p. In addition, LINC01234 knockdown inhibited the tumorigenesis of liver cancer via inactivating TGF-β signaling. Furthermore, silencing of LINC01234 notably inhibited the tumor growth of liver cancer in vivo.

Conclusion

Downregulation of LINC01234 could inhibit the tumorigenesis of liver cancer via mediation of miR-513a-5p/USP4/TGF-β axis. Thus, LINC01234 might serve as a new target for the treatment of liver cancer.

A Comprehensive Narrative Review on the Evolving Role of Endoscopic Ultrasound in Focal Solid Liver Lesions Diagnosis and Management

The implications of endoscopic ultrasound (EUS) have expanded considerably in recent years to cover more fields in invasive gastroenterology practice, as both an investigative and therapeutic modality. The utility of EUS in the diagnosis and management of focal liver lesions has gained a special attractiveness recently. The EUS probe proximity to the liver and its excellent spatial resolution enables real-time images coupled with several enhancement techniques, such as contrast-enhanced (CE) EUS. Aside from its notable capability to execute targeted biopsies and therapeutic interventions, EUS has developed into a hopeful therapeutic tool for the management of solid liver lesions. Herein, we provide a comprehensive state-of-the-art review on the efficacy and safety of EUS in the diagnosis and management of focal solid liver lesions. Medline/PubMed and Embase database searches were conducted by two separate authors (T.K. and W.S.), all relevant studies were assessed, and relevant data was extracted and fully reported. EUS-guided diagnosis of focal liver lesions by sonographic morphologic appearance and cytological and histopathological finding of biopsies obtained via fine needle aspiration/biopsy have been shown to significantly improve the diagnosis of solid liver lesions compared with traditional imaging tools. Similarly, EUS-guided treatment has been shown to consistently have excellent technical success, high efficacy, and minor adverse events. The evolving valuable evidences of EUS utility might satisfy the unmet need of optimizing management of focal solid liver lesions.