Computer-aided Classification of Liver Lesions from CT Images Based on Multiple ROI

Introduction: It is not uncommon for clinicians to see focal liver lesions (FLLs), which can be anything from harmless hemangiomas to dangerous hepatocellular carcinoma (HCC) and metastases. In order to properly treat these lesions, an accurate diagnosis is required. The purpose of this research is to compare magnetic resonance imaging (MRI) results with histological ones in order to determine how well MRI can detect and characterize localized liver lesions. Materials and Methods: This prospective study comprised 50 patients with localized liver lesions identified via ultrasonography or computed tomography (CT) and then assessed using MRI. This study was conducted at the Department of Radiology at Gouri Devi Institute of Medical Science and Hospital, Rajbandh, West Bengal, India from June 2021 to May 2022. Patients received MRI images with T1-weighted, T2-weighted, diffusion-weighted imaging (DWI), and contrast-enhanced dynamic imaging with gadolinium-based contrast agents. The imaging findings were analyzed according to lesion morphology, signal intensity, enhancement patterns, and diffusion restriction. All patients received biopsy or surgical resection, with histological findings serving as the reference standard. The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of MRI in identifying malignant and benign liver lesions were assessed. Results: There were 30 cancerous tumors and 20 benign ones among the 50 individuals. To distinguish between malignant and benign lesions, MRI showed a sensitivity of 94%, specificity of 89%, PPV of 93%, and NPV of 90%. Malignant lesions affecting the liver were most commonly found as hepatocellular carcinoma (HCC) and metastatic liver lesions, respectively. In terms of benign lesions, hemangiomas, hepatic adenomas, and focal nodular hyperplasia (FNH) were detected most often. Even while most benign lesions showed facilitated diffusion on diffusion-weighted imaging (DWI), 85% of malignant lesions showed restricted diffusion. Contrast-enhanced sequences showed that malignant lesions had arterial-phase hyperenhancement with washout, but benign lesions showed distinctive enhancement patterns as homogenous enhancement in FNH and peripheral nodular enhancement in hemangiomas. Conclusion: Magnetic resonance imaging (MRI) provides a very accurate picture of localized liver lesions and correlates very well with histological results. Diagnostic certainty in distinguishing benign from malignant liver lesions is greatly increased when morphologic evaluation, diffusion restriction, and dynamic contrast enhancement patterns are combined. To better characterize liver lesions, MRI should be regarded the imaging modality of choice. This will help reduce the need for biopsies that aren't essential and guide appropriate clinical therapy.

Deep learning-based diagnosis systems are useful to identify abnormalities in medical images with the greatly increased workload of doctors. Specifically, the rate of new cases and deaths from malignancies is rising for liver diseases. Early detection of liver lesions plays an extremely important role in effective treatment and gives a higher chance of survival for patients. Therefore, automatic detection and classification of common liver lesions are essential for doctors. In fact, radiologists mainly rely on Hounsfield Units to locate liver lesions but previous studies often pay little attention to this factor. In this paper, we propose an improved method for the automatic classification of common liver lesions based on deep learning techniques and the variation of Hounsfield Unit densities on CT images with and without contrast. Hounsfield Unit is used to locate liver lesions accurately and support data labeling for classification. We construct a multi-phase classification model developed on the deep neural networks of Faster R-CNN, R-FCN, SSD, and Mask R-CNN with the transfer learning approach. The experiments are conducted on six scenarios with multi-phase CT images of common liver lesions. Experimental results show that the proposed method improves the detection and classification of liver lesions compared with recent methods because its accuracy achieves up to 97.4%. The proposed models are very useful to assist doctors in the automatic segmentation and classification of liver lesions to solve the problem of depending on the clinician's experience in the diagnosis and treatment of liver lesions.

To analyse characteristics of benign and malignant liver tumours in dynamic and static MR imaging with the superparamagnetic MR contrast medium Resovist. All 30 patients were examined on a 1.5 Tesla MR unit (Magnetom 63 SP, Siemens AG, Erlangen, Germany) using proton density (PD) weighted (w), T2-weighted-spin-echo, a T1-weighted SE, and a T1-weighted FLASH-2 D gradient echo (GRE) sequence before, during and after the application of Resovist. Dynamic imaging was performed using a T2-weighted GRE-sequence (TurboFLASH; TR/TE = 11/30; flip angle = 10 degrees; Tl = 600 ms). Histopathology revealed benign liver lesions in 8 patients and malignant lesions in 22 patients. Dynamic T2-weighted sequence revealed an early loss of signal intensity in normal liver parenchyma (percentage signal intensity loss (PSIL) = 40.0 +/- 12.2% by 4 mumol Fe/kg, 47.2 +/- 18.8% by 8 mumol Fe/kg and 62.7 +/- 13.0% by 16 mumol Fe/kg), in the spleen, as well as in FNH (PSIL = 49.5 +/- 7.3% by 8 mumol Fe/kg), and regenerating nodules in the first minute after application of Resovist. In two of 4 cases with HCC a short drop in signal intensity was immediately observed after the application, whereas signal intensity remained unchanged in all other malignant liver tumours. Enhanced PDw and T2-weighted SE-sequences revealed an improved detection and delineation of malignant liver lesions versus plain MR imaging. 17 liver lesions of a size lower 10 mm were additionally detected in postcontrast T2-weighted SE-sequences in 4 patients. Dynamic and static versus plain MR imaging of primary and secondary liver lesions is markedly improved by the superparamagnetic contrast material Resovist, especially in case of intravenous bolus application of this liver-specific contrast medium.

With the widespread use of cross-sectional imaging (US, CT, MRI) the detection of unsuspected liver lesions, asymptomatic prior to their discovery, has become common. These lesions may become a source of anxiety and often require further investigation to reassure the patient of their usually benign nature. However, in some cases unsuspected lesions could be malignant like small primary hepatocellular carcinoma or metastases from previously undiagnosed primary cancer. Furthermore it is well known that benign and malignant lesions may coexist. Thus the appropriate way to address these lesions correctly is a growing clinical problem. Of course it is important to recognize that the range of diagnostic findings encountered will be related to the population under study. While patients with benign diseases appendicitis, cholecystitis ect. are likely to have benign liver lesions, patients with known primary cancer will have a greater likelihood of findings consistent with malignant hepatic lesions. Proper use of modern hepato-biliary imaging modalities, such as dynamic CE-CT, CE-US, and MRI, in combination with simple laboratory tests often allow a definite diagnosis to be made without resorting to exhaustive investigation or inappropriate biopsy or surgery. The optimal approach to evaluate incidental hepatic lesions may be the combination of the appropriate imaging technique, i.e. high-quality MRI, and an experienced radioloist. During this lecture we will discuss the problem of incidentally detected liver lesions and develop a framework of further imaging studies to use for the diagnosis of these lesions. Typically clinical situations will be demonstrated and the value of the different imaging modalities, especially high quality MRI will be shown.

Introduction: Diffusion-weighted imaging [DWI] plays a significant role in distinguishing liver lesions. This study focuses on the significance of apparent diffusion coefficient [ADC] values in distinguishing between benign and malignant liver lesions.
 Methods: A retrospective evaluation was conducted on patients who underwent liver MRI examinations at our clinic, and 378 liver lesions were found. The study comprised 141 women, 80 men, and 221 patients with liver lesions. The measured ADC values between benign and malignant lesions were compared. In addition, the lesion ADC value/CSF ADC value ratio was measured, and the relative ADC value was named.
 Results: 118 hemangiomas, 59 simple cysts, 67 hydatid cysts, 3 focal nodular hyperplasias, 9 abscesses, 28 hepatocellular carcinomas, 88 metastases, and 6 lymphomas were evaluated. Simple cysts and hemangiomas had the most elevated ADC values compared to all other liver lesions. The ADC values for liver lesions classified as benign had a mean of 2.10 x 10-3 mm2/s, while malignant ones had a lower mean of 0.75 x 10-3 mm2/s. The ADC values of malignant liver lesions were lower compared to the benign lesions.
 Conclusion: Using DWI combined with ADC values could be valuable in discerning between benign and malignant liver lesions. Besides, relative ADC values may also contribute to more objective results.

Diffusion weighted imaging of liver lesions suspect for metastases: Apparent diffusion coefficient (ADC) values and lesion contrast are independent from Gd-EOB-DTPA administration

Summary Introduction. The liver is a parenchimal abdominal organ with wide variety of primary benign or malignant tumors as well as metastatic tumors. Aim of the study. Was to evaluate the informativity and usefulness of abdominal ultrasound (US) and computer tomography (CT) imaging results in diagnostics of focal liver lesions. Material and methods. The study was conducted in Pauls Stradins Clinical University Hospital. Retrospective analysis of 126 patients with focal liver lesions was performed in the time period of 5 years (2009 till 2014). The medical records of patients with focal liver lesions were analyzed. Primary diagnosis detected by US or CT was evaluated and compared with final morphology after surgical operation or liver biopsy. The obtained results were expressed in percent and analyzed. Sensitivity and specificity of CT scan to detect malignant hepatic lesions was estimated and expressed as percentage with 95% confidence interval. Results. A total of 126 patients with diagnosed liver lesions were included in the study. 96 patients were in the group with performed liver CT scans, with median age of 58.9 years. 30 patients were included in the group with liver US, with median age of 60.1 years. Liver biopsy under US control was performed for 95 patients, but surgical operations with liver resections for 31 patients. 86 patients had malignant liver lesions, but 40 had benign liver lesions. In the US group primary and final diagnosis agreement was in 26 (87%) cases, but diagnosis disagreement in 4 (13%) cases. Diagnosis disagreement was found in 26 (27.1%) cases in the CT group, but agreement was in 70 (72.9%) cases. Overall sensitivity of CT to detect malignant hepatic lesions was 95.2% (95%CI 86.7-98.3%) and specificity was 64.7% (95%CI 47.9-78.5%). Conclusions. CT is a good imaging method for detection of focal liver lesions. In case of unclear diagnosis, percutaneous liver puncture biopsy is recommended. It is mandatory to develop a unified CT scan and US investigation protocol to improve the quality of investigation as well as further treatment tactics.

4071 Background: In patients with colorectal cancer detection and differentiation of liver metastases is important for further treatment decissions. Recently diffusion weighted MR imaging (DWI) was introduced as an new technique for liver tumor detection and characterzation. DWI is sensitive to molecular diffusion which is the random thermal motion of molecules (Brownian motion). MR imaging is the only method by which molecular diffusion processes can be evaluated in vivo. New techniques, such as breath-hold imaging or respiratory triggered single shot echo-planar-imaging (EPI) allow to extent DWI to abdominal organs, especially the liver. First results show that DWI can be used for the reliable detection of small subcentimeter liver lesions. We evaluated this new imaging method with regard to liver lesion detection and lesion characterization in patients with colorectal cancer. Methods: 115 consecutive patients with colorectal cancer and suspected liver metastases were evaluated with multi slice-CT and MRI at 1.5 Tesla using T1- and T2-weighted pulse sequences as well as DWI with respiratory triggered single shot echo-planar-imaging (EPI). Images were analyzed by two experienced radiologist blinded to the clinical results. Results for lesion detection for CT and DWI-MRI were compared using ROC analysis. Furthermore, in a subset of 60 patients diffusion coefficients (ADCs) were calculated for the differentiation of liver lesions (benign vs malignant). Results: For lesion detection ROC analysis showed superiority (p< 0.05) of DWI as compared to MS-CT. Especially metastases with diameters of smaller 1 cm were better detected with DWI-MRI as compared to MS-CT (p < 0.02). Results for differentiation show mean ADC-values of 1.24 for normal liver while malignant tumors have ADCs of 1.04–1.22 and benign lesions such as hemangiomas or cysts of 1.9 to 3.0, respectively. Furthermore, therapeutic changes were made in 20 % of patients based on DWI. Conclusions: MR imaging using DWI is more sensitive than multi- detector row CT in the detection and characterization of liver metatstases from colorectal cancer. Thus, it can be used as a new method for liver lesion detection and characterization as well as for optimal treatment planning. No significant financial relationships to disclose.

The objective of our study was to evaluate a new 3D fast spoiled gradient-recalled echo (FSPGR) sequence referred to as modified liver acceleration volume acquisition (LAVA) for high-resolution gadolinium-enhanced dual arterial phase liver MRI and to determine the effect of this technique on the timing of the contrast bolus and lesion detection. Gadolinium-enhanced dual arterial phase liver MRI was performed in 109 patients using a modified LAVA sequence that supports adaptive 2D centric view ordering, efficient 2D autocalibrated acceleration, and partial-Fourier to achieve faster scan times while maintaining the same slice thickness, resolution, and coverage as single-phase imaging. After a fixed 20-second scan delay, a modified LAVA acquisition required a single 24- to 26-second breath-hold for two arterial phases with 56-60 slices per pass. Images were reviewed for timing relative to liver enhancement, lesion conspicuity, and lesion detection. Liver lesion depiction was evaluated qualitatively and quantitatively. A control group of 109 patients underwent imaging using a single arterial phase 3D FSPGR sequence, which was also performed with a fixed 20-second scan delay. The single arterial phase images produced optimal timing in the middle or late arterial phase in 79 (72%) of the 109 control group patients compared with 99 (91%) of the 109 study group patients who underwent imaging using a dynamic modified LAVA dual arterial phase sequence. For the modified LAVA sequence, the first-pass images were obtained during the mid arterial phase in 34 patients (31%). The second-pass images were obtained during the mid arterial phase in 51 patients (47%) and late arterial phase in 26 patients (24%). Sixty-two patients had liver lesions showing greater conspicuity--on the first phase in 17 patients (27%) and second phase in 45 patients (73%). Hypovascular lesions were more conspicuous on second-phase images in 24 (86%) of 28 patients. Hypervascular lesions were more conspicuous on first-phase images in 13 patients (38%) and on second-phase images in 21 (62%) of 34 patients. The first-phase images detected 165 and 155 liver lesions, respectively, for two observers compared with 233 and 224 lesions on the second-phase images, whereas the combined dual arterial phase images detected 256 and 248 hepatic lesions. High-resolution dual arterial phase 3D FSPGR MRI improves the timing of the arterial phase of liver enhancement and provides additional information for liver lesion detection.

Purpose. To assess the diagnostic accuracy of dynamic susceptibility contrast-enhanced magnetic resonance imaging (DSCE-MRI) in differentiation between benign and malignant liver lesions by assessment of tumoral perfusion parameters. Methods Materials. Seventy-three patients with known focal liver lesions, including 45 benign (16 FNH, 27 angiomas, and 2 abscesses) and 28 malignant ones (17 metastases, 9 HCCs, and 2 cholangiocarcinoma) underwent 1.5 T MRI upper abdominal study, with standard protocol that included dynamic contrast-enhanced sequences. On dedicated workstation, time-intensity curves were determined and the following perfusion parameters were calculated: relative arterial, venous and late enhancement (RAE, RVE, RLE), maximum enhancement (ME), relative enhancement (RE), and time to peak (TTP). Results. All diagnoses were established either by histopathology or imaging follow-up. Perfusion mean values calculated in benign lesions were RAE 33.8%, RVE 66.03%, RLE 80.63%, ME 776.00%, MRE 86.27%, and TTP 146.95 sec. Corresponding perfusion values calculated in malignant lesions were RAE 22.47%, RVE 40.54%, RLE 47.52%, ME 448.78%, MRE 49.85%, and TTP 183.79 sec. Statistical difference (p < 0.05) was achieved in all the perfusion parameters calculated, obtaining different cluster of perfusion kinetics between benign and malignant lesions. Conclusions. DSCE-MRI depicts kinetic differences in perfusion parameters among the different common liver lesions, related to tumour supply and microvascular characteristics.

Perfusion CT of the liver typically involves scanning the liver at least 20 times, resulting in a large radiation dose. We developed and validated a simplified model of tumor blood supply that can be applied to standard triphasic scans and evaluated whether this can be used to distinguish benign and malignant liver lesions. Triphasic CTs of 46 malignant and 32 benign liver lesions were analyzed. For each phase, regions of interest were drawn in the arterially enhancing portion of each lesion, as well as the background liver, aorta, and portal vein. Hepatic artery and portal vein blood supply coefficients for each lesion were then calculated by expressing the enhancement curve of the lesion as a linear combination of the enhancement curves of the aorta and portal vein. Hepatocellular carcinoma (HCC) and hypervascular metastases, on average, both had increased hepatic artery coefficients compared to the background liver. Compared to HCC, benign lesions, on average, had either a greater hepatic artery coefficient (hemangioma) or a greater portal vein coefficient (focal nodular hyperplasia or transient hepatic attenuation difference). Hypervascularity with washout is a key diagnostic criterion for HCC, but it had a sensitivity of 72% and specificity of 81% for diagnosing malignancy in our diverse set of liver lesions. The sensitivity for malignancy was increased to 89% by including enhancing lesions that were hypodense on all phases. The specificity for malignancy was increased to 97% (p = 0.039) by also examining hepatic artery and portal vein blood supply coefficients, while maintaining a sensitivity of 76%.

In this study we investigated the sensitivity of high resolution ultrasound (HRU) in the detection of small liver tumors and its microcirculation in a humanized tumor mouse model (HTM). These mice develop a complete human immune system and human breast cancer growth in the liver which allows the investigation of antibody based immunotherapies under human like conditions. HTM were generated by the co-transplantation of human breast cancer cells and human hematopoietic stem cells. HRU, Doppler sonography (CCDS), contrast enhanced ultrasound (CEUS) and color-coded elastography were performed on all HTM and confirmed by histopathological assessment. Using HRU and CEUS, noncystic solid liver lesions between 2 and 11 mm (mean 3.5 mm) size were detectable in HTM. Granulomatous areas were identified by B-scan imaging, showing areas of higher stiffness in elastography and areas without contrast media uptake in the late phase (CEUS). In addition, CEUS detected capillary microcirculation of benign and malignant liver lesions smaller than 10 mm. Beyond human breast cancer HTM additionally developed small parenchymal liver lesions, which could be characterized by HRU in combination with CEUS and elastography in-vivo. Nevertheless, the defined diagnoses of solid liver lesions less than 5 mm require confirmation by histopathology.

In patients with known malignancy, correct detection and characterization of liver lesions has important therapeutic consequences. Conventional sonography is the most commonly used modality for liver imaging in tumor patients. However, it has a lower sensitivity for the detection of liver metastases compared to contrast-enhanced computed tomography (CT) and magnetic resonance imaging (MRI). The majority of liver metastases are hypoechoic and well defined in baseline ultrasound (US), while detection of isoechoic or small liver metastases <1 cm is difficult and the differentiation of liver metastases from benign liver lesions and other malignant liver tumors can be impossible with baseline US. The use of microbubble-based ultrasound contrast agents and contrast-specific imaging techniques advanced the accuracy of ultrasound in liver imaging. Levovist and SonoVue are the US contrast agents approved for liver imaging in Europe. Compared to Levovist, SonoVue allows continuous imaging of the liver in real-time over a period of up to 5 minutes. As a result, SonoVue became the preferred contrast agent for liver imaging in the recent years, while Levovist became less important. Important for the detection of liver metastases are the portal venous and late phases in which metastases show a wash-out and can be detected as hypoechoic lesions in homogeneous enhanced liver parenchyma. The detection of hepatic metastases is substantially improved by CEUS compared to conventional B-mode sonography. Several studies showed sensitivity in detecting liver metastases comparable to that of contrast-enhanced CT and MRI. Furthermore, the typical enhancement patterns of the different benign and malignant liver lesions allow reliable characterization and differentiation from liver metastases in the majority of cases. This paper provides information about the advantages and expedient application of contrast-enhanced ultrasound (CEUS) in tumor patients.

Objective To discuss the value of conventional ultrasound combined with virtual touch tissue imaging (VTI) in differentiating benign and malignant breast lesions by breast imaging reporting and data system (BI-RADS) diagnostic criteria. Methods From January 2013 to October 2014, 395 breast lesions from 389 patients from the Tenth People's Hospital, Tongji University were included in present study. All the lesions were confirmed by surgery pathology. Conventional ultrasound and VTI examinations were done before surgery. Both benign and malignant lesions were classified into varied categories based on the boundary of lesions under VTI mode. Benign and malignant breast lesions were classified into BI-RADS categories by conventional ultrasound or conventional ultrasound combined with VTI. The χ2 test was used to compare the differences of lesions′ boundaries between VTI categories and VTI mode. The pathology findings were used as golden criteria, to calculate the sensitivity, specificity and accuracy of conventional ultrasound as well as conventional ultrasound combined with VTI. And their diagnostic performances were compared by McNemar′s test. Results Of 395 breast lesions, 282 (71.4%) were benign, 113 (28.6%) were malignant. 75.5% (213/282) benign lesions were classified as Ⅱ-Ⅲ under VTI mode, while 83.2% (94/113) malignant lesions were classified as Ⅳ-Ⅴ under VTI mode. The differences in classification between them was statistically significant (χ2=114.759, P<0.01). 77.7% (219/282) benign lesions had unclear boundaries under VTI mode, while 78.8% (89/113) malignant lesions had clear boundaries under VTI mode. The differences in boundaries between them was statistically significant (χ2=108.480, P<0.01). The sensitivity, specificity and accuracy of conventional ultrasound in differentiating benign and malignant breast lesions were 97.3%, 54.2% and 66.6% respectively; while the sensitivity, specificity and accuracy of conventional ultrasound combined with virtual touch tissue imaging in differentiating benign and malignant breast lesions were 98.2%, 84.4% and 88.4% respectively. The specificity and accuracy were higher when only using conventional ultrasound with statistical significance (χ2=79.425, 67.549, both P<0.01); however the sensitivity had no significantly different (χ2=11.974, P<0.01). Conclusion Conventional ultrasound combined with virtual touch tissue imaging can improve the specificity and the accuracy in differentiating benign and malignant breast lesions. Key words: Elasticity imaging techniques; Ultrasonography; Breast diseases; Diagnosis, differential

The use of contrast enhancers has widened the possibilities of sonographic imaging, and allows the differentiation of characteristic enhancement patterns leading to diagnosis in focal liver lesions. The aim of our study was to evaluate contrast ultrasound signs in diagnosing malignant liver lesions. 86 patients with 100 solid liver lesions were enrolled. A baseline gray-scale sonogram was obtained with a multifrequency 4 C convex array probe, followed by contrast-enhanced sonography with a low mechanical index (<0.2) over 300 seconds. Final diagnosis was confirmed by histology or in case of haemangioma by CT/NMR and quantitative contrast harmonic imaging (CHI) with perfusion analysis (contrast). 55 malignant (6 HCC, 46 secondary malignant lesions - 3 of them lymphoma, 3 cholangiocarcinoma), and 45 benign lesions (8 FNH, 1 von Meyenburg complex, 1 granuloma, 3 adenoma, 21 hemangioma, 2 focal fat storage imbalances, 7 abscesses, one scar, and in one case normal liver) were found. 51/55 malignant (all but one filia and three HCC), but also 17/45 benign lesions showed hypoperfusion in the late phase. The ultrasound pattern in the arterial phase differed in malignant lesions: 22 lesions were initially hypervascular, 20 had rim enhancement and in 13 lesions there was a non-specific vascularisation. In all but one malignant lesion a diminishing of contrast agent in the late phase compared to the arterial phase with respect to the surrounding liver tissue was observed. Only three benign lesions with this later sign were falsely diagnosed as malignant: one adenoma, one epitheloid granuloma, and a scar. Quantitative perfusion pattern was analyzed exemplary. Diminishing of contrast agent in the late phase compared to the arterial phase with respect to the surrounding liver tissue as a sign for malignancy had a positive predictive value of 95%, a sensitivity of 98%, a negative predictive value of 98%, and a specificity of 93%. Diminishing of contrast agent in the late phase compared to the arterial phase with respect to the surrounding liver tissue is a helpful sign in contrast enhanced ultrasound to diagnose malignancies.