Malignancy diagnosis of liver lesion in contrast enhanced ultrasound using an end-to-end method based on deep learning.

BACKGROUND: Contrast-enhanced ultrasound (CEUS) is considered as an efficient tool for focal liver lesion characterization, given it allows real-time scanning and provides dynamic tissue perfusion information. An accurate diagnosis of liver lesions with CEUS requires a precise interpretation of CEUS images. However,it is a highly experience dependent task which requires amount of training and practice. To help improve the constrains, this study aims to develop an end-to-end method based on deep learning to make malignancy diagnosis of liver lesions using CEUS. METHODS: A total of 420 focal liver lesions with 136 benign cases and 284 malignant cases were included. A deep learning model based on a two-dimensional convolution neural network, a long short-term memory (LSTM), and a linear classifier (with sigmoid) was developed to analyze the CEUS loops from different contrast imaging phases. For comparison, a 3D-CNN based method and a machine-learning (ML)-based time-intensity curve (TIC) method were also implemented for performance evaluation. RESULTS: Results of the 4-fold validation demonstrate that the mean AUC is 0.91, 0.88, and 0.78 for the proposed method, the 3D-CNN based method, and the ML-based TIC method, respectively. CONCLUSIONS: The proposed CNN-LSTM method is promising in making malignancy diagnosis of liver lesions in CEUS without any additional manual features selection.

Deep Multi-Objective Learning from Low-Dose CT for Automatic Lung-RADS Report Generation.

Radiology report generation through chest radiography interpretation is a time-consuming task that involves the interpretation of images by expert radiologists. It is common for fatigue-induced diagnostic error to occur, and especially difficult in areas of the world where radiologists are not available or lack diagnostic expertise. In this research, we proposed a multi-objective deep learning model called CT2Rep (Computed Tomography to Report) for generating lung radiology reports by extracting semantic features from lung CT scans. A total of 458 CT scans were used in this research, from which 107 radiomics features and 6 slices of segmentation related nodule features were extracted for the input of our model. The CT2Rep can simultaneously predict position, margin, and texture, which are three important indicators of lung cancer, and achieves remarkable performance with an F1-score of 87.29%. We conducted a satisfaction survey for estimating the practicality of CT2Rep, and the results show that 95% of the reports received satisfactory ratings. The results demonstrate the great potential in this model for the production of robust and reliable quantitative lung diagnosis reports. Medical personnel can obtain important indicators simply by providing the lung CT scan to the system, which can bring about the widespread application of the proposed framework.

Ultrasonic Microelastography to Assess Biomechanical Properties of the Cornea.

OBJECTIVE: To both qualitatively and quantitatively investigate corneal biomechanical properties through an ultrasonic microelastography imaging system, which is potentially useful in the diagnosis of diseases, such as keratoconus, postrefractive keratectasia, and tracking treatment such as cross-linking surgery. METHODS: Our imaging system has a dual-frequency configuration, including a 4.5 MHz ring transducer to push the tissue and a confocally aligned 40 MHz needle transducer to track micron-level displacement. Two-dimensional/three-dimensional acoustic radiation force impulse (ARFI) imaging and Young's modulus in the region of interest were performed on ex vivo porcine corneas that were either cross-linked using formalin solution or preloaded with intraocular pressure (IOPs) from 5 to 30 mmHg. RESULTS: The increase of corneal stiffness and the change in cross-linked volume following formalin crosslinking could be precisely observed in the ARFI images and reflected by the reconstructed Young's modulus while the B-mode structural images remained almost unchanged. In addition, the relationship between the stiffness of the cornea and IOPs was investigated among 12 porcine corneas. The corneal stiffness is significantly different at various IOPs and has a tendency to become stiffer with increasing IOP. CONCLUSION: Our results demonstrate the principle of using ultrasonic microelastography techniques to image the biomechanical properties of the cornea. Integrating high-resolution ARFI imaging labeled with reconstructed Young's modulus and structural imaging of the cornea can potentially lead to a routinely performed imaging modality in the field of ophthalmology.

A Novel Adhesion Index for Verifying the Extent of Adhesion for the Extensor Digitorum Communis in Patients with Metacarpal Fractures.

This study aims to determine if the relative displacement between the extensor digitorum communis (EDC) tendon and its surrounding tissues can be used as an adhesion index (AI) for assessing adhesion in metacarpal fractures by comparing two clinical measures, namely single-digit-force and extensor lag (i.e., the difference between passive extension and full active extension). The Fisher-Tippett block-matching method and a Kalman-filter algorithm were used to determine the relative displacements in 39 healthy subjects and 8 patients with metacarpal fractures. A goniometer was used to measure the extensor lag, and a force sensor was used to measure the single-digit-force. Measurements were obtained twice for each patient to evaluate the performance of the AI in assessing the progress of rehabilitation. The Pearson correlation coefficient was calculated to quantify the various correlations between the AI, extensor lag, and single-digit-force. The results showed strong correlations between the AI and the extensor lag, the AI and the single-digit-force, and the extensor lag and the single-digit-force (r = 0.718, -0.849, and -0.741; P = 0.002, P < 0.001, and P = 0.001, respectively). The AI in the patients gradually decreased after continuous rehabilitation, but remained higher than that of healthy participants.

Evaluating the intensity of the acoustic radiation force impulse (ARFI) in intravascular ultrasound (IVUS) imaging: Preliminary in vitro results.

The ability to measure the elastic properties of plaques and vessels is significant in clinical diagnosis, particularly for detecting a vulnerable plaque. A novel concept of combining intravascular ultrasound (IVUS) imaging and acoustic radiation force impulse (ARFI) imaging has recently been proposed. This method has potential in elastography for distinguishing between the stiffness of plaques and arterial vessel walls. However, the intensity of the acoustic radiation force requires calibration as a standard for the further development of an ARFI-IVUS imaging device that could be used in clinical applications. In this study, a dual-frequency transducer with 11MHz and 48MHz was used to measure the association between the biological tissue displacement and the applied acoustic radiation force. The output intensity of the acoustic radiation force generated by the pushing element ranged from 1.8 to 57.9mW/cm(2), as measured using a calibrated hydrophone. The results reveal that all of the acoustic intensities produced by the transducer in the experiments were within the limits specified by FDA regulations and could still displace the biological tissues. Furthermore, blood clots with different hematocrits, which have elastic properties similar to the lipid pool of plaques, with stiffness ranging from 0.5 to 1.9kPa could be displaced from 1 to 4µm, whereas the porcine arteries with stiffness ranging from 120 to 291kPa were displaced from 0.4 to 1.3µm when an acoustic intensity of 57.9mW/cm(2) was used. The in vitro ARFI images of the artery with a blood clot and artificial arteriosclerosis showed a clear distinction of the stiffness distributions of the vessel wall. All the results reveal that ARFI-IVUS imaging has the potential to distinguish the elastic properties of plaques and vessels. Moreover, the acoustic intensity used in ARFI imaging has been experimentally quantified. Although the size of this two-element transducer is unsuitable for IVUS imaging, the experimental results reported herein can be applied in ARFI-IVUS imaging applications.

High-resolution acoustic-radiation-force-impulse imaging for assessing corneal sclerosis.

In ophthalmology, detecting the biomechanical properties of the cornea can provide valuable information about various corneal pathologies, including keratoconus and the phototoxic effects of ultraviolet radiation on the cornea. Also, the mechanical properties of the cornea can be used to evaluate the recovery from corneal refractive surgeries. Therefore, noninvasive and high-resolution estimation of the stiffness distribution in the cornea is important in ophthalmic diagnosis. The present study established a method for high-resolution acoustic-radiation-force-impulse (ARFI) imaging based on a dual-frequency confocal transducer in order to obtain a relative stiffness map, which was used to assess corneal sclerosis. An 11-MHz pushing element was used to induce localized displacements of tissue, which were monitored by a 48-MHz imaging element. Since the tissue displacements are directly correlated with the tissue elastic properties, the stiffness distribution in a tiny region of the cornea can be found by a mechanical B/D scan. The experimental system was verified using tissue-mimicking phantoms that included different geometric structures. Ex vivo cornea experiments were carried out using fresh porcine eyeballs. Corneas with localized sclerosis were created artificially by the injection of a formalin solution. The phantom experiments showed that the distributions of stiffness within different phantoms can be recognized clearly using ARFI imaging, and the measured lateral and axial resolutions of this imaging system were 177 and 153 µ m, respectively. The ex vivo experimental results from ARFI imaging showed that a tiny region of localized sclerosis in the cornea could be distinguished. All of the obtained results demonstrate that high-resolution ARFI imaging has considerable potential for the clinical diagnosis of corneal sclerosis.