Detection Of Small Lesions Research Articles

Breast Lesion Detection for Ultrasound Images Using MaskFormer.

This study evaluates the performance of the MaskFormer model for segmenting and classifying breast lesions using ultrasound images, addressing ultrasound's limitations. Ultrasound used for breast cancer detection faces challenges like low image contrast and difficulty in the detection of small or multiple lesions, further complicated by variability based on operator skill. Initial experiments with U-Net and other CNN-based models revealed constraints, such as early plateauing in model loss, which indicated suboptimal learning and performance. In contrast, MaskFormer demonstrated continuous improvement, achieving higher precision in breast lesion segmentation and significantly reducing both false positives and false negatives. Comparative analysis showed MaskFormer's superior performance, with the highest precision and recall rates for malignant lesions and an overall mean average precision (mAP) of 0.943. The model's ability to detect a diverse range of breast lesions, including those potentially missed by the human eye, especially by less experienced practitioners, underscores its potential. These findings suggest that integrating AI models like MaskFormer could greatly enhance ultrasound performance for breast cancer detection, providing reliable, operator-independent image analysis and potentially improving patient outcomes on a global scale.

Photon-counting computed tomography versus energy-integrating computed tomography for detection of small liver lesions: comparison using a virtual framework imaging.

Photon-counting computed tomography (PCCT) has the potential to provide superior image quality to energy-integrating CT (EICT). We objectively compare PCCT to EICT for liver lesion detection. Fifty anthropomorphic, computational phantoms with inserted liver lesions were generated. Contrast-enhanced scans of each phantom were simulated at the portal venous phase. The acquisitions were done using DukeSim, a validated CT simulation platform. Scans were simulated at two dose levels ( 1.5 to 6.0mGy) modeling PCCT (NAEOTOM Alpha, Siemens, Erlangen, Germany) and EICT (SOMATOM Flash, Siemens). Images were reconstructed with varying levels of kernel sharpness (soft, medium, sharp). To provide a quantitative estimate of image quality, the modulation transfer function (MTF), frequency at 50% of the MTF ( ), noise magnitude, contrast-to-noise ratio (CNR, per lesion), and detectability index ( , per lesion) were measured. Across all studied conditions, the best detection performance, measured by , was for PCCT images with the highest dose level and softest kernel. With soft kernel reconstruction, PCCT demonstrated improved lesion CNR and compared with EICT, with a mean increase in CNR of 35.0% ( ) and 21% ( ) and a mean increase in of 41.0% ( ) and 23.3% ( ) for the 1.5 and 6.0mGy acquisitions, respectively. The improvements were greatest for larger phantoms, low-contrast lesions, and low-dose scans. PCCT demonstrated objective improvement in liver lesion detection and image quality metrics compared with EICT. These advances may lead to earlier and more accurate liver lesion detection, thus improving patient care.

Assessing the deep learning based image quality enhancements for the BGO based GE omni legend PET/CT

BackgroundThis study investigates the integration of Artificial Intelligence (AI) in compensating the lack of time-of-flight (TOF) of the GE Omni Legend PET/CT, which utilizes BGO scintillation crystals.MethodsThe current study evaluates the image quality of the GE Omni Legend PET/CT using a NEMA IQ phantom. It investigates the impact on imaging performance of various deep learning precision levels (low, medium, high) across different data acquisition durations. Quantitative analysis was performed using metrics such as contrast recovery coefficient (CRC), background variability (BV), and contrast to noise Ratio (CNR). Additionally, patient images reconstructed with various deep learning precision levels are presented to illustrate the impact on image quality.ResultsThe deep learning approach significantly reduced background variability, particularly for the smallest region of interest. We observed improvements in background variability of 11.8%\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\%$$\\end{document}, 17.2%\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\%$$\\end{document}, and 14.3%\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\%$$\\end{document} for low, medium, and high precision deep learning, respectively. The results also indicate a significant improvement in larger spheres when considering both background variability and contrast recovery coefficient. The high precision deep learning approach proved advantageous for short scans and exhibited potential in improving detectability of small lesions. The exemplary patient study shows that the noise was suppressed for all deep learning cases, but low precision deep learning also reduced the lesion contrast (about −30%\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\%$$\\end{document}), while high precision deep learning increased the contrast (about 10%\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\%$$\\end{document}).ConclusionThis study conducted a thorough evaluation of deep learning algorithms in the GE Omni Legend PET/CT scanner, demonstrating that these methods enhance image quality, with notable improvements in CRC and CNR, thereby optimizing lesion detectability and offering opportunities to reduce image acquisition time.

Abdominal Applications of Photon-Counting CT.

Photon-counting computed tomography (PCCT) has shown promising advancements in abdominal imaging in clinical use. Though more peer-reviewed primary literature is needed, this commentary explores PCCT's potential applications, focusing on enhancing diagnostic accuracy, optimizing radiation dose management, and improving patient care. PCCT offers improved spatial and contrast resolution, lower image noise, and reduced radiation dose. Increased spatial resolution provides better detail in abdominal imaging, aiding in the detection of small lesions and subtle pathological changes. However, this generates more images per scan, raising concerns about "image overload" in PACS, potentially leading to longer reading times and increased stress for radiologists. PCCT's improved contrast resolution enhances tissue differentiation, reducing the need for intravenous contrast agents. The technology's advanced tissue characterization provides several advantages, such as non-invasive and opportunistic liver disease evaluation and improved differentiation of renal and adrenal masses. PCCT's optimized radiation dose management is crucial for patients requiring frequent scans. Enhanced diagnostic accuracy through spectral information aids in tissue differentiation, improving confidence in diagnoses. Streamlined workflows, particularly in emergency settings, and oncologic imaging, are potential benefits, reducing the need for additional imaging studies. Future integration of PCCT into clinical practice requires collaboration, education, and research to fully harness its potential, ensuring optimized abdominal imaging and improved patient care.

Endoscopic Ultrasound and Intraductal Ultrasound in the Diagnosis of Biliary Tract Diseases: A Narrative Review.

Endoscopic ultrasound (EUS) and intraductal ultrasound (IDUS) play very important roles in the field of biliary tract disease. Because of their excellent spatial resolution, the detection of small lesions and T- or N-staging of tumors have become possible. Additionally, contrast-enhanced EUS and the new imaging technique of detective flow imaging are reported to be useful for differential diagnosis. Furthermore, EUS-guided tissue acquisition is used not only for pathological diagnosis but also to collect tissue samples for cancer genome profiling. This review provides an overview of diagnosis utilizing the features and techniques of EUS and IDUS.

Relevance of lesion size in navigator-triggered and free-breathing diffusion-weighted liver MRI.

The purpose of this study was to investigate the relevance of focal liver lesions (FLL) size for lesion detection comparing navigator triggering (TRIG) to free breathing (FB) liver Diffusion-weighted magnetic resonance imaging (DWI). Patients with known or suspected FLL were prospectively (registry number 276_19 B) included from October to December 2019 in this study, out of which 32 had liver lesions. Echo planar spin-echo DWI data both with TRIG and FB were with approximately constant acquisition times acquired at 1.5 T. Lesions were segmented in the b = 800 s/mm² images in both the TRIG and FB images. The lesion size, location (liver segment), liver lesion visibility, as well as contrast-to-noise ratio (CNR) were recorded. The CNR was assessed with the Wilcoxon-Mann-Whitney test and the number of visible lesions with the Fisher test. Data from 43 patients (22 female) were analyzed. The mean patient age was 58 ± 14 years. A total of 885 FLL (Ntotal) were segmented. Among these, 811 lesions (Nboth) were detected with TRIG and FB, 65 lesions exclusively with TRIG (NTRIG_Only), and nine exclusively in FB (NFB_Only). The largest additional lesion in TRIG/FB had a diameter of 10.4 mm/7.6 mm. The number of additional lesions detected with TRIG decreased with size. Among all lesions ≤ 4.7 mm, the relative number of additional lesions was 15.6%. Additional lesions were found in all liver segments with TRIG. In the left liver lobe, the relative proportion was 9.2%, and in the right liver lobe 5.4%. CNR and visibility were significantly higher in TRIG than in FB (p < 0.001). In relation to size, the difference is significant in terms of visibility and CNR for lesion diameters ≤ 8 mm. Respiration triggering can improve the detection of small liver lesions with diameters up to approx. 1 cm in the whole liver. Question Can respiration triggering (TRIG) improve the detection of small FLL compared to FB diffusion-weighted imaging? Findings Among 885 segmented FLL, TRIG was superior to FB for lesions smaller than 8 mm and had improved CNR and visibility. Clinical relevance Diffusion-weighted magnetic resonance imaging is used for the detection of focal liver lesions and image quality is influenced by breathing motion. Navigator triggering becomes more important for smaller lesions, and seems recommendable for the detection of small focal liver lesions.

Mammography classification with multi-view deep learning techniques: Investigating graph and transformer-based architectures

The potential and promise of deep learning systems to provide an independent assessment and relieve radiologists’ burden in screening mammography have been recognized in several studies. However, the low cancer prevalence, the need to process high-resolution images, and the need to combine information from multiple views and scales still pose technical challenges. Multi-view architectures that combine information from the four mammographic views to produce an exam-level classification score are a promising approach to the automated processing of screening mammography. However, training such architectures from exam-level labels, without relying on pixel-level supervision, requires very large datasets and may result in suboptimal accuracy. Emerging architectures such as Visual Transformers (ViT) and graph-based architectures can potentially integrate ipsi-lateral and contra-lateral breast views better than traditional convolutional neural networks, thanks to their stronger ability of modeling long-range dependencies. In this paper, we extensively evaluate novel transformer-based and graph-based architectures against state-of-the-art multi-view convolutional neural networks, trained in a weakly-supervised setting on a middle-scale dataset, both in terms of performance and interpretability. Extensive experiments on the CSAW dataset suggest that, while transformer-based architecture outperform other architectures, different inductive biases lead to complementary strengths and weaknesses, as each architecture is sensitive to different signs and mammographic features. Hence, an ensemble of different architectures should be preferred over a winner-takes-all approach to achieve more accurate and robust results. Overall, the findings highlight the potential of a wide range of multi-view architectures for breast cancer classification, even in datasets of relatively modest size, although the detection of small lesions remains challenging without pixel-wise supervision or ad-hoc networks.

Value of hepatic artery digital subtraction angiography in the detection of small hepatocellular carcinoma lesions.

To compare the detection rates of hepatic artery digital subtraction angiography (HA-DSA) and magnetic resonance imaging (MRI) gadolinium diethylenetriaminepentaacetic acid (MRI-Gd-DTPA) and MRI gadolinium diethylenetriaminepentaacetic acid (MRI-Gd-EOB-DTPA) for small (diameter ≤2 cm) hepatocellular carcinoma (HCC) lesions. A prospective analysis of patients admitted to the Tumor Hospital of Guangxi Medical University between January 1, 2015, and December 30, 2016, was conducted. The detection rates of the three methods were analyzed. The diameter of small HCC lesions detected using HA-DSA and MRI-Gd-EOB-DTPA were evaluated. The diagnostic value of HCC Barcelona staging for HA-DSA was analyzed. For 107 small lesions detected in 57 patients, the detection rates of HA-DSA and MRI-Gd-DTPA were 86.0% (92/107) and 71.0% (76/107), respectively (p < .05). Of 77 small lesions detected in 42 patients using MRI-Gd-EOB-DTPA and HA-DSA, 67 were detected using HA-DSA, all of which had a rich blood supply, and 72 were detected using MRI-Gd-EOB-DTPA. The minimum diameter of lesions detected using MRI-Gd-EOB-DTPA was approximately 0.4 cm, whereas that of lesions detected using HA-DSA was approximately 0.5 cm. After HA-DSA, a change in the Barcelona staging occurred in 33.3% (62/186) of cases but not after MRI-Gd-DTPA; HA-DSA was significantly better than MRI-Gd-DTPA for staging (p = .03). HA-DSA and MRI-Gd-EOB-DTPA have high diagnostic values for the detection of small HCC lesions, which is helpful for accurate staging of HCC and provides the most valuable information for patient treatment and prognosis.

YOLOv5s-BiPCNeXt, a Lightweight Model for Detecting Disease in Eggplant Leaves.

Ensuring the healthy growth of eggplants requires the precise detection of leaf diseases, which can significantly boost yield and economic income. Improving the efficiency of plant disease identification in natural scenes is currently a crucial issue. This study aims to provide an efficient detection method suitable for disease detection in natural scenes. A lightweight detection model, YOLOv5s-BiPCNeXt, is proposed. This model utilizes the MobileNeXt backbone to reduce network parameters and computational complexity and includes a lightweight C3-BiPC neck module. Additionally, a multi-scale cross-spatial attention mechanism (EMA) is integrated into the neck network, and the nearest neighbor interpolation algorithm is replaced with the content-aware feature recombination operator (CARAFE), enhancing the model's ability to perceive multidimensional information and extract multiscale disease features and improving the spatial resolution of the disease feature map. These improvements enhance the detection accuracy for eggplant leaves, effectively reducing missed and incorrect detections caused by complex backgrounds and improving the detection and localization of small lesions at the early stages of brown spot and powdery mildew diseases. Experimental results show that the YOLOv5s-BiPCNeXt model achieves an average precision (AP) of 94.9% for brown spot disease, 95.0% for powdery mildew, and 99.5% for healthy leaves. Deployed on a Jetson Orin Nano edge detection device, the model attains an average recognition speed of 26 FPS (Frame Per Second), meeting real-time requirements. Compared to other algorithms, YOLOv5s-BiPCNeXt demonstrates superior overall performance, accurately detecting plant diseases under natural conditions and offering valuable technical support for the prevention and treatment of eggplant leaf diseases.

Single-operator single-balloon enteroscopy is safe and effective: A 6-year retrospective study.

Single-balloon enteroscopy (SBE) is an effective tool for the detection of small intestine lesions. Because itis conventionally performed by two operators, the efficacy of single-operator SBE method has not yet been elucidated. We aimed to evaluate the diagnostic yield, total enteroscopy rate, procedure time, and complications of single-operator SBE for small intestinal disease. This was a single-center, retrospective study including consecutive patients who underwent single-operator SBE for suspicious small intestinal disorders or required therapeutic interventions between December 2014 and January 2019. The SBE procedures were performed by four endoscopists. Diagnostic yield, total enteroscopy rate, procedure time, incubation depth, and complications were analyzed, and stratification analysis was performed. Altogether 922 patients with 1422 SBE procedures were included for analysis, among whom 250, 172, and 500 patients underwent SBE via the oral route, the anal route and a combined route, respectively. The overall diagnostic yield was 78.52% (724/922). And 253 patients achieved total enteroscopy, with a total enteroscopy rate of 56.10%. The average procedure time for the oral and anal routes were 69.28 ± 14.72 min and 64.95 ± 13.87 min, respectively. While the incubation depth was 389.95 ± 131.42 cm and 191.81 ± 83.67 cm, respectively. Jejunal perforation was observed in one patient, which was managed by endoclips. Stratification analysis showed that the diagnostic yield and total enteroscopy rate significantly increased with operation experience together with decreased procedure time. Single-operator SBE is effective and safe for the detection of small intestinal lesions, and is easy to master.

Deep learning based bilateral filtering for edge-preserving denoising of respiratory-gated PET

BackgroundResidual image noise is substantial in positron emission tomography (PET) and one of the factors limiting lesion detection, quantification, and overall image quality. Thus, improving noise reduction remains of considerable interest. This is especially true for respiratory-gated PET investigations. The only broadly used approach for noise reduction in PET imaging has been the application of low-pass filters, usually Gaussians, which however leads to loss of spatial resolution and increased partial volume effects affecting detectability of small lesions and quantitative data evaluation. The bilateral filter (BF) — a locally adaptive image filter — allows to reduce image noise while preserving well defined object edges but manual optimization of the filter parameters for a given PET scan can be tedious and time-consuming, hampering its clinical use. In this work we have investigated to what extent a suitable deep learning based approach can resolve this issue by training a suitable network with the target of reproducing the results of manually adjusted case-specific bilateral filtering.MethodsAltogether, 69 respiratory-gated clinical PET/CT scans with three different tracers ([18F]\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$[^{18}\ ext {F}]$$\\end{document}FDG, [18F]\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$[^{18}\ ext {F}]$$\\end{document}L-DOPA, [68Ga]\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$[^{68}\ ext {Ga}]$$\\end{document}DOTATATE) were used for the present investigation. Prior to data processing, the gated data sets were split, resulting in a total of 552 single-gate image volumes. For each of these image volumes, four 3D ROIs were delineated: one ROI for image noise assessment and three ROIs for focal uptake (e.g. tumor lesions) measurements at different target/background contrast levels. An automated procedure was used to perform a brute force search of the two-dimensional BF parameter space for each data set to identify the “optimal” filter parameters to generate user-approved ground truth input data consisting of pairs of original and optimally BF filtered images. For reproducing the optimal BF filtering, we employed a modified 3D U-Net CNN incorporating residual learning principle. The network training and evaluation was performed using a 5-fold cross-validation scheme. The influence of filtering on lesion SUV quantification and image noise level was assessed by calculating absolute and fractional differences between the CNN, manual BF, or original (STD) data sets in the previously defined ROIs.ResultsThe automated procedure used for filter parameter determination chose adequate filter parameters for the majority of the data sets with only 19 patient data sets requiring manual tuning. Evaluation of the focal uptake ROIs revealed that CNN as well as BF based filtering essentially maintain the focal SUVmax\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\ ext {SUV}_\ ext {max}$$\\end{document} values of the unfiltered images with a low mean ± SD difference of δSUVmaxCNN,STD\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\delta \ ext {SUV}_\ ext {max}^{\ ext {CNN},\ ext {STD}}$$\\end{document} = (−3.9 ± 5.2)% and δSUVmaxBF,STD\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\delta \ ext {SUV}_\ ext {max}^{\ ext {BF},\ ext {STD}}$$\\end{document} = (−4.4 ± 5.3)%. Regarding relative performance of CNN versus BF, both methods lead to very similar SUVmax\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\ ext {SUV}_\ ext {max}$$\\end{document} values in the vast majority of cases with an overall average difference of δSUVmaxCNN,BF\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\delta \ ext {SUV}_\ ext {max}^{\ ext {CNN},\ ext {BF}}$$\\end{document} = (0.5 ± 4.8)%. Evaluation of the noise properties showed that CNN filtering mostly satisfactorily reproduces the noise level and characteristics of BF with δNoiseCNN,BF\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\delta \ ext {Noise}^{\ ext {CNN},\ ext {BF}}$$\\end{document} = (5.6 ± 10.5)%. No significant tracer dependent differences between CNN and BF were observed.ConclusionsOur results show that a neural network based denoising can reproduce the results of a case by case optimized BF in a fully automated way. Apart from rare cases it led to images of practically identical quality regarding noise level, edge preservation, and signal recovery. We believe such a network might proof especially useful in the context of improved motion correction of respiratory-gated PET studies but could also help to establish BF-equivalent edge-preserving CNN filtering in clinical PET since it obviates time consuming manual BF parameter tuning.

Leveraging a realistic synthetic database to learn Shape-from-Shading for estimating the colon depth in colonoscopy images

Colonoscopy is the choice procedure to diagnose, screening, and treat the colon and rectum cancer, from early detection of small precancerous lesions (polyps), to confirmation of malign masses. However, the high variability of the organ appearance and the complex shape of both the colon wall and structures of interest make this exploration difficult. Learned visuospatial and perceptual abilities mitigate technical limitations in clinical practice by proper estimation of the intestinal depth. This work introduces a novel methodology to estimate colon depth maps in single frames from monocular colonoscopy videos. The generated depth map is inferred from the shading variation of the colon wall with respect to the light source, as learned from a realistic synthetic database. Briefly, a classic convolutional neural network architecture is trained from scratch to estimate the depth map, improving sharp depth estimations in haustral folds and polyps by a custom loss function that minimizes the estimation error in edges and curvatures. The network was trained by a custom synthetic colonoscopy database herein constructed and released, composed of 248400 frames (47 videos), with depth annotations at the level of pixels. This collection comprehends 5 subsets of videos with progressively higher levels of visual complexity. Evaluation of the depth estimation with the synthetic database reached a threshold accuracy of 95.65%, and a mean-RMSE of 0.451cm, while a qualitative assessment with a real database showed consistent depth estimations, visually evaluated by the expert gastroenterologist coauthoring this paper. Finally, the method achieved competitive performance with respect to another state-of-the-art method using a public synthetic database and comparable results in a set of images with other five state-of-the-art methods. Additionally, three-dimensional reconstructions demonstrated useful approximations of the gastrointestinal tract geometry. Code for reproducing the reported results and the dataset are available at https://github.com/Cimalab-unal/ColonDepthEstimation.

Revisiting the "Puffed Cheek" Technique: Advantages, Fallacies, and Potential Solutions.

The "puffed cheek" technique is routinely performed during CT neck studies in patients with suspected oral cavity cancers. The insufflation of air within the oral vestibule helps in the detection of small buccal mucosal lesions, with better delineation of lesion origin, depth, and extent of spread. The pitfalls associated with this technique are often underrecognized and poorly understood. They can mimic actual lesions, forfeiting the technique's primary purpose. This review provides an overview of the puffed cheek technique and its associated pitfalls. These pitfalls include pneumoparotid, soft palate elevation that resembles a nasopharyngeal mass, various tongue displacements or distortions that obscure tongue lesions or mimic them, sublingual gland herniation, an apparent exacerbation of the airway edema, vocal cord adduction that hinders glottic evaluation, and false indications of osteochondronecrosis in laryngeal cartilage. Most stem from a common underlying mechanism of unintentional Valsalva maneuver engaged in by the patient while trying to perform a puffed cheek, creating a closed air column under positive pressure with resultant surrounding soft-tissue displacement. These pitfalls can thus be avoided by instructing the patient to maintain continuous nasal breathing while puffing out their cheek during image acquisition, preventing the formation of the closed air column. Keywords: CT, Head/Neck © RSNA, 2024.

Harmonisation of quantitative assessment between different generation of PET/CT: Biograph mCT and Biograph Vision

The usage of modern positron emission tomography scanners, in particular with digital detectors, allows obtaining images with better quality, increases the detection of small pathological lesions, reduces scanning time and the activity administered to the patient which leads to a decrease of patient dose as well. However, the values of the quantitative image parameters shift upward, which can lead to significant differences with the quantitative assessment obtained on the previous generation device. In order to compare quantitative assessments obtained on different generations of PET/CT, it is necessary to harmonise quantitative image parameters and perform regular quality control. The aim of current work is the comparison of different methods for harmonization of quantitative image parameters on the example of harmonisation of two PET/CT: Biograph mCT 128 and Biograph Vision 600. NEMA IEC Body phantom filled with 18F solution was scanned in Listmode in two bed positions with overlap in the sphere area during five minutes per bed position. Recovery coefficient used for harmonisation was measured for each sphere of the phantom. Harmonisation between Vision and mCT was performed using two methods: choosing of harmonised reconstruction parameters and EQ.PET technology. The acceptable divergence range between the recovery coefficients for Vision and for mCT is ±10% (20% range). The recovery coefficients measured for reconstruction: 4 iterations and 5 subsets, ToF+PSF, Gaussian 7 mm, matrix 220x220 completely fit within the 20% range. The recovery coefficients measured using EQ = 6 mm (optimal value) fit within the 20% range except the spheres with a diameter of 10 and 13 mm. Both harmonisation methods allow to approximate the quantitative assessment/ However, EQ.PET has limitations for the small lesions. Choosing harmonised reconstruction parameters is the mostwidely used harmonisation method; the EQ.PET allows to harmonise quantitative assessment without the use of multiple reconstruction protocols and losses in visualization ability

Same-day post-therapy imaging with a new generation whole-body digital SPECT/CT in assessing treatment response to [177Lu]Lu-PSMA-617 in metastatic castration-resistant prostate cancer.

Lutetium-177 [177Lu]Lu-PSMA-617 radioligand therapy (RLT) represents a significant advancement for metastatic castration-resistant prostate cancer (mCRPC), demonstrating improvements in radiographic progression free survival (rPFS) and overall survival (OS) with a low rate of associated side effects. Currently, most post-therapy SPECT/CT is conducted at 24h after infusion. This study examines the clinical utility of a next-generation multi-detector Cadmium-Zinc-Telluride (CZT) SPECT/CT system (StarGuide) in same-day post-infusion assessment and early treatment response to [177Lu]Lu-PSMA-617. In this retrospective study, 68 men with progressive mCRPC treated with [177Lu]Lu-PSMA-617 at our center from June 2022 to June 2023 were evaluated. Digital whole-body SPECT/CT imaging was performed after [177Lu]Lu-PSMA-617infusion (mean ± SD: 1.8 ± 0.6h, range 1.1-4.9h). Quantitative analysis of [177Lu]Lu-PSMA-617 positive lesions was performed in patients who underwent at least 2 post-therapy SPECT/CT, using liver parenchyma uptake as reference. Metrics including [177Lu]Lu-PSMA-617 positive total tumor volume (Lu-TTV), SUVmax and SUVmean were calculated. These quantitative metrics on post-infusion SPECT/CT images after cycles 1, 2 and 3 were correlated with overall survival (OS), prostate specific antigen-progression free survival (PSA-PFS) as defined by prostate cancer working group 3 (PCWG3), and PSA decrease over 50% (PSA50) response rates. 56 patients (means age 76.2 ± 8.1 years, range: 60-93) who underwent at least 2 post-therapy SPECT/CT were included in the image analysis. The whole-body SPECT/CT scans (~ 12min per scan) were well tolerated, with 221 same-day scans performed (89%). At a median of 10-months follow-up, 33 (58.9%) patients achieved PSA50 after [177Lu]Lu-PSMA-617 treatment and median PSA-PFS was 5.0 months (range: 1.0-15 months) while median OS was not reached. Quantitative analysis of SPECT/CT images showed that 37 patients (66%) had > 30% reduction in Lu-TTV, associated with significantly improved overall survival (median not reached vs. 6 months, P = 0.008) and PSA-PFS (median 6 months vs. 1 months, P < 0.001). However, changes in SUVmax or SUVmean did not correlate with PSA-PFS or OS. We successfully implemented same-day post-therapy SPECT/CT after [177Lu]Lu-PSMA-617 infusions. Quantitation of 1-2h post-therapy SPECT/CT images is a promising method for assessing treatment response. However, the approach is currently limited by its suboptimal detection of small tumor lesions and the necessity of incorporating a third-cycle SPECT/CT to mitigate the effects of any potential treatment-related flare-up. Further investigation in a larger patient cohort and prospective validation is essential to confirm these findings and to explore the role of SPECT/CT as a potential adjunct to PSMA PET/CT in managing mCRPC.

Automated detection of small bowel lesions based on capsule endoscopy using deep learning algorithm

BackgroundIn order to overcome the challenges of lesion detection in capsule endoscopy (CE), we improved the YOLOv5-based deep learning algorithm and established the CE-YOLOv5 algorithm to identify small bowel lesions captured by CE. MethodsA total of 124,678 typical abnormal images from 1,452 patients were enrolled to train the CE-YOLOv5 model. Then 298 patients with suspected small bowel lesions detected by CE were prospectively enrolled in the testing phase of the study. Small bowel images and videos from the above 298 patients were interpreted by the experts, non-experts and CE-YOLOv5, respectively. ResultsThe sensitivity of CE-YOLOv5 in diagnosing vascular lesions, ulcerated/erosive lesions, protruding lesions, parasite, diverticulum, active bleeding and villous lesions based on CE videos was 91.9 %, 92.2 %, 91.4 %, 93.1 %, 93.3 %, 95.1 %, and 100 % respectively. Furthermore, CE-YOLOv5 achieved specificity and accuracy of more than 90 % for all lesions. Compared with experts, the CE-YOLOv5 showed comparable overall sensitivity, specificity and accuracy (all P > 0.05). Compared with non-experts, the CE-YOLOv5 showed significantly higher overall sensitivity (P < 0.0001) and overall accuracy (P < 0.0001), and a moderately higher overall specificity (P = 0.0351). Furthermore, the time for AI-reading (5.62 ± 2.81 min) was significantly shorter than that for the other two groups (both P < 0.0001). ConclusionsCE-YOLOv5 diagnosed small bowel lesions in CE videos with high sensitivity, specificity and accuracy, providing a reliable approach for automated lesion detection in real-world clinical practice.

Endoscopic loco-regional treatment in controlling pancreatic neuroendocrine tumors (PNETs) behavior: a case series and literature review.

Pancreatic neuroendocrine tumors (PNETs) are considered rare pancreatic neoplasms, and it is a challenging disease entity due to its indolent behavior and is difficult to manage. Diagnostic challenge is usually found in the imaging-based approach, such as transabdominal ultrasound, abdominal CT scan, and abdominal MRI. Surgery is still the main key player in controlling the disease. The main problems in clinical practice are the early detection of small PNETs lesion and non-functional PNET (NF-PNET) cases. Most cases usually come with large tumor size or metastatic disease. Endoscopic ultrasound (EUS) has been well-known as the most sensitive tool for early detection of pancreatic malignancy. It has now also been developed for managing pancreatic cancer, such as tumor ablation therapy. We presented four variety of PNETs cases (one patient with non-functioning PNETs and three patients with functioning PNETs) who have been successfully treated with EUS-guided radiofrequency ablation (RFA) with good control of tumor growth during follow-up. One patient with a large-size insulinoma could have the tumor controlled gradually after several EUS-RFA sessions. No adverse events or major complications were observed during and after the EUS-RFA procedure. EUS has been shown as a promising tool not only for more accurate diagnosis but also for tumor growth control. However, it would need further comparison studies between EUS and surgical approaches to make a better treatment strategy.

Image quality evaluation for a clinical organ-targeted PET camera.

A newly developed clinical organ-targeted Positron Emission Tomography (PET) system (also known as Radialis PET) is tested with a set of standardized and custom tests previously used to evaluate the performance of Positron Emission Mammography (PEM) systems. Imaging characteristics impacting standardized uptake value (SUV) and detectability of small lesions, namely spatial resolution, linearity, uniformity, and recovery coefficients, are evaluated. In-plane spatial resolution was measured as 2.3 mm ± 0.1 mm, spatial accuracy was 0.1 mm, and uniformity measured with flood field and NEMA NU-4 phantom was 11.7% and 8.3% respectively. Selected clinical images are provided as reference to the imaging capabilities under different clinical conditions such as reduced activity of 2-[fluorine-18]-fluoro-2-deoxy-D-glucose (18F-FDG) and time-delayed acquisitions. SUV measurements were performed for selected clinical acquisitions to demonstrate a capability for quantitative image assessment of different types of cancer including for invasive lobular carcinoma with comparatively low metabolic activity. Quantitative imaging performance assessment with phantoms demonstrates improved contrast recovery and spill-over ratio for this PET technology when compared to other commercial organ-dedicated PET systems with similar spatial resolution. Recovery coefficients were measured to be 0.21 for the 1 mm hot rod and up to 0.89 for the 5 mm hot rod of NEMA NU-4 Image Quality phantom. Demonstrated ability to accurately reconstruct activity in tumors as small as 5 mm suggests that the Radialis PET technology may be well suited for emerging clinical applications such as image guided assessment of response to neoadjuvant systemic treatment (NST) in lesions smaller than 2 cm. Also, our results suggest that, while spatial resolution greatly influences the partial volume effect which degrades contrast recovery, optimized count rate performance and image reconstruction workflow may improve recovery coefficients for systems with comparable spatial resolution. We emphasize that recovery coefficient should be considered as a primary performance metric when a PET system is used for accurate lesion size or radiotracer uptake assessments.

Multiparametric magnetic resonance imaging for hepatocellular carcinoma, part1 : Morphology and dynamic perfusion imaging in primary diagnostics and treatment monitoring

Radiology plays akey role in the diagnosis and monitoring of hepatocellular carcinoma (HCC). Ultrasound, computed tomography (CT) and magnetic resonance imaging (MRI) are used to identify HCC lesions. Multiparametric MRI provides detailed insights into the tumor biology through the analysis of morphology, perfusion and diffusion. In this way preoperative decisions can be optimized. The guidelines recommend using contrast-enhanced MRI or ultrasound for the diagnosis of HCC. The preferred method is MRI due to its superiority in the detection of small lesions The treatment response is evaluated using modified response evaluation criteria for solid tumors (RECIST) and the European Association for the Study of the Liver (EASL) criteria. The use of multiparametric MRI in conjunction with the liver imaging reporting and data system (LI-RADS) plays overall acentral role in the precise diagnosis and monitoring of the treatment of HCC.

Enhancing gadoxetic acid–enhanced liver MRI: a synergistic approach with deep learning CAIPIRINHA-VIBE and optimized fat suppression techniques

ObjectiveTo investigate whether a deep learning (DL) controlled aliasing in parallel imaging results in higher acceleration (CAIPIRINHA)-volumetric interpolated breath-hold examination (VIBE) technique can improve image quality, lesion conspicuity, and lesion detection compared to a standard CAIPIRINHA-VIBE technique in gadoxetic acid–enhanced liver MRI.MethodsThis retrospective single-center study included 168 patients who underwent gadoxetic acid–enhanced liver MRI at 3 T using both standard CAIPIRINHA-VIBE and DL CAIPIRINHA-VIBE techniques on pre-contrast and hepatobiliary phase (HBP) images. Additionally, high-resolution (HR) DL CAIPIRINHA-VIBE was obtained with 1-mm slice thickness on the HBP. Three abdominal radiologists independently assessed the image quality and lesion conspicuity of pre-contrast and HBP images. Statistical analyses involved the Wilcoxon signed-rank test for image quality assessment and the generalized estimation equation for lesion conspicuity and detection evaluation.ResultsDL and HR-DL CAIPIRINHA-VIBE demonstrated significantly improved overall image quality and reduced artifacts on pre-contrast and HBP images compared to standard CAIPIRINHA-VIBE (p < 0.001), with a shorter acquisition time (DL vs standard, 11 s vs 17 s). However, the former presented a more synthetic appearance (both p < 0.05). HR-DL CAIPIRINHA-VIBE showed superior lesion conspicuity to standard and DL CAIPIRINHA-VIBE on HBP images (p < 0.001). Moreover, HR-DL CAIPIRINHA-VIBE exhibited a significantly higher detection rate of small (< 2 cm) solid focal liver lesions (FLLs) on HBP images compared to standard CAIPIRINHA-VIBE (92.5% vs 87.4%; odds ratio = 1.83; p = 0.036).ConclusionDL and HR-DL CAIPIRINHA-VIBE achieved superior image quality compared to standard CAIPIRINHA-VIBE. Additionally, HR-DL CAIPIRINHA-VIBE improved the lesion conspicuity and detection of small solid FLLs. DL and HR-DL CAIPIRINHA-VIBE hold the potential clinical utility for gadoxetic acid–enhanced liver MRI.Clinical relevance statementDL and HR-DL CAIPIRINHA-VIBE hold promise as potential alternatives to standard CAIPIRINHA-VIBE in routine clinical liver MRI, improving the image quality and lesion conspicuity, enhancing the detection of small (< 2 cm) solid focal liver lesions, and reducing the acquisition time.Key Points• DL and HR-DL CAIPIRINHA-VIBE demonstrated improved overall image quality and reduced artifacts on pre-contrast and HBP images compared to standard CAIPIRINHA-VIBE, in addition to a shorter acquisition time.• DL and HR-DL CAIPIRINHA-VIBE yielded a more synthetic appearance than standard CAIPIRINHA-VIBE.• HR-DL CAIPIRINHA-VIBE showed improved lesion conspicuity than standard CAIPIRINHA-VIBE on HBP images, with a higher detection of small (< 2 cm) solid focal liver lesions.