Search

The standard assessment of response to cancer treatments is based on gross tumor characteristics, such as tumor size or glycolysis, which provide very indirect information about the effect of precision treatments on the pharmacological targets of tumors. Several advanced imaging modalities allow for the visualization of targeted tumor hallmarks. Descriptors extracted from these images can help establishing new classifications of precision treatment response. We propose a machine learning (ML) framework to analyze metabolic–anatomical–vascular imaging features from positron emission tomography, ultrafast Doppler, and computed tomography in a mouse model of paraganglioma undergoing anti-angiogenic treatment with sunitinib. Imaging features from the follow-up of sunitinib-treated (n= 8, imaged once-per-week/6-weeks) and sham-treated (n= 8, imaged once-per-week/3-weeks) mice groups were dimensionally reduced and analyzed with hierarchical clustering Analysis (HCA). The classes extracted from HCA were used with 10 ML classifiers to find a generalized tumor stage prediction model, which was validated with an independent dataset of sunitinib-treated mice. HCA provided three stages of treatment response that were validated using the best-performing ML classifier. The Gaussian naive Bayes classifier showed the best performance, with a training accuracy of 98.7 and an average area under curve of 100. Our results show that metabolic–anatomical–vascular markers allow defining treatment response trajectories that reflect the efficacy of an anti-angiogenic drug on the tumor target hallmark.

Purpose: We present a markerless vision-based method for on-the-fly three-dimensional (3D) pose estimation of a fiberscope instrument to target pathologic areas in the endoscopic view during exploration. Approach: A 2.5-mm-diameter fiberscope is inserted through the endoscope’s operating channel and connected to an additional camera to perform complementary observation of a targeted area such as a multimodal magnifier. The 3D pose of the fiberscope is estimated frame-by-frame by maximizing the similarity between its silhouette (automatically detected in the endoscopic view using a deep learning neural network) and a cylindrical shape bound to a kinematic model reduced to three degrees-of-freedom. An alignment of the cylinder axis, based on Plücker coordinates from the straight edges detected in the image, makes convergence faster and more reliable. Results: The performance on simulations has been validated with a virtual trajectory mimicking endoscopic exploration and on real images of a chessboard pattern acquired with different endoscopic configurations. The experiments demonstrated a good accuracy and robustness of the proposed algorithm with errors of [Formula: see text] in distance position and [Formula: see text] in axis orientation for the 3D pose estimation, which reveals its superiority over previous approaches. This allows multimodal image registration with sufficient accuracy of [Formula: see text]. Conclusion: Our pose estimation pipeline was executed on simulations and patterns; the results demonstrate the robustness of our method and the potential of fiber-optical instrument image-based tracking for pose estimation and multimodal registration. It can be fully implemented in software and therefore easily integrated into a routine clinical environment.

The backscatter coefficient (BSC) quantifies the frequency-dependent reflectivity of tissues. Accurate estimation of the BSC requires knowledge of the attenuation coefficient slope (ACS) of tissues in the beam path between the transducer and the insonified region of interest, namely, the total attenuation. In this study, the total attenuation is calculated as the cumulative sum of values of a local attenuation map devised using full angular spatial compounding (FASC). The BSC was parameterized through the integrated backscatter coefficient (iBSC) obtaining iBSC maps. Experimental validation of the proposed approach consisted of scanning two cylindrical physical phantoms with off-centered inclusions having different ACS and BSC values than the background. Additional iBSC maps were computed assuming an uniform ACS map of 0.5 dB/cm/MHz (which is a value assumed for soft tissues) instead of the FASC-ACS map. Finally a iBSC map was obtained using an ideal ACS map formed with ground truth ACS values and knowledge of inclusion true position. The results were comparable when using the FASC-ACS map or the ideal ACS map in term of inclusion detectability and estimation accuracy. The use of the uniform ACS map resulted in some cases with very high fractional error (>;9 dB), which highlights the relevance of accurate compensation for total attenuation. These results suggest that BSCs can be reliably estimated using total attenuation compensation from FASC-ACS maps.

The backscatter coefficient (BSC) quantifies the frequency-dependent reflectivity of tissues. Accurate estimation of the BSC is only possible with the knowledge of the attenuation coefficient slope (ACS) of the tissues under examination. In this study, the use of attenuation maps constructed using full angular spatial compounding (FASC) is proposed for attenuation compensation when imaging integrated BSCs. Experimental validation of the proposed approach was obtained using two cylindrical physical phantoms with off-centered inclusions having different ACS and BSC values than the background, and in a phantom containing an ex vivo chicken breast sample embedded in an agar matrix. With the phantom data, three different ACS maps were employed for attenuation compensation: (1) a ground truth ACS map constructed using insertion loss techniques, (2) the estimated ACS map using FASC attenuation imaging, and (3) a uniform ACS map with a value of 0.5 dBcm\protect \relax \special {t4ht=−}1MHz\protect \relax \special {t4ht=−}1, which is commonly used to represent attenuation in soft tissues. Comparable results were obtained when using the ground truth and FASC-estimated ACS maps in term of inclusion detectability and estimation accuracy, with averaged fractional error below 2.8 dB in both phantoms. Conversely, the use of the homogeneous ACS map resulted in higher levels of fractional error ( > 10 dB), which demonstrates the importance of an accurate attenuation compensation. The results with the ex vivo tissue sample were consistent with the observations using the physical phantoms, with the FASC-derived ACS map providing comparable BSC images to those formed using the ground truth ACS map and more accurate than those BSC images formed using a uniform ACS. These results suggest that BSCs can be reliably estimated using FASC when a self-consistent attenuation compensation stemming from prior estimation of an accurate ACS map is used.

Cavity volume is an important clinical index for the assessment of the healing process and effectiveness of treatment applied on chronic ulcers. Recently, 3D scanners have proven to effectively track ulcer’s volume evolution. However, photogrammetry presents itself as a low cost and portable alternative. We conducted a comparative study between photogrammetric and 3D scanner-based volume estimation of small skin ulcers. A total of 24 cutaneous leishmaniasis ulcers’ virtual models were generated using a commercial laser scanner and a full-HD portable camera. The reconstruction from videos was performed using open-source software (VisualSFM). The results revealed a similar performance when compared to 3D Scan-based volume estimation with a median deviation and percentile error of 33.68 mm3 and 21.10%, respectively. In addition, the method proved to effectively follow the same tendency in the assessment of healing ulcers when compared to the 3D-scanner. Moreover, when comparing the two virtual models trough residual differences after ICP registration the mean error and standard deviation was 0.36 ± 0.27 mm.

Cutaneous Leishmaniasis is a skin infection which is commonly present in underdeveloped countries. The incidence is particularly high in amazonic countries of Latin-America like Brazil, Colombia and Perú and is usually reported as endemic. In Perú, more than one million people are at risk of infection and approximately 6,000 new cases are detected each year. The present work proposes to integrate a set of control, monitoring and disease quantification procedures in: (1) an automated tool to expedite the analysis in laboratories studying parasiticidal agents and (2) a non-invasive treatment monitoring protocol. The first, consists in the adaptation of an optical microscope KRUSS MBL3100 to perform a fast image capture and automated promastigotes identification. This may be of value to evaluate the effectiveness of various parasiticidal agents. The counting process is performed by an automatic segmentation in the RGB green color space discriminating elements by their area. The second, proposes a protocol for monitoring the evolution of the disease treatment divided into three stages: supra-skin modeling and reconstruction, subcutaneous exploration by textural characteristics and volumetric segmentation. This protocol is performed using a Next Engine HD 3D scanner and a Vevo Visualsonix 2100 ultrasonic scanner. The results show improvements in sample processing time, accuracy and inter- and intra-operational variability. The sensitivity and accuracy of the microscopic identification system was of 97% and 92% respectively. The exactitude and precision error was up to 2% and the sensitivity and specificity went as high as 71% in the 3D reconstruction and texture analysis respectively.

This paper presents a landmark-free approach to estimate the fiberscope pose during endoscopic exploration for in-vivo optical biopsy. The fiberscope pose is estimated by fitting the projection of a virtual 3D cylinder into the endoscopic images. The cylinder axis is estimated based on the apparent contours using Plucker coordinates and its insertion is estimated by maximizing the similarity between binary masks. The performance of the method is evaluated on simulations: the mean Euclidian distance of fiberscopic tip between estimated pose and ground truth is 0.158 ± 0.113 mm. The in-vivo performance is assessed in two endoscopic sequences by comparing automatic RCF and manual segmentations in terms of angular deviation of the axis and Euclidian distance between the tip location. The estimation of the relative position of both cameras allows to perform registration between the two image modalities.

International audience; We developed a multispectral-augmented endoscopic prototype which increases the common number of bands under analysis, allowing exploration in the visible and near infrared range (400–1000 nm). The prototype combines endoscopy with spectroscopy using white light (WL) or Narrow Band Imaging light (NBI) endoscope and two multispectral cameras connected to a twin-cam splitter. The splitter is then connected to an optical fiber and introduced in the endoscope instrument channel.In this work, we introduce a spectral calibration and an axial displacement correction function to register both modalities. The former is based on a multi-linear transformation of multispectral bands and its performance is assessed using a Digital SG ColorChecker® pattern to report an RMSE of 6.78%. The latter relates the insertion depth of the fiberscope with the required geometric transformation. The performance was assessed using a chessboard pattern and its corner coordinates as ground truth. The mean RMSE error for the registration using our method was 2.3 ± 0.7 pixels, whereas the RMSE error using a frame by frame homographic registration was 1.2 ± 0.4 pixels. Finally, the technique was tested on mouth exploration samples to simulate in-vivo acquisition and display complete spectra for single points of analysis. The results reveal that our method provides similar performance when compared to a homographic transformation which would be impossible to perform in-vivo.

Medical ultrasound has proven to be superior to other imaging modalities while scanning underserved areas. Compared to the alternatives, it is fast, safe, portable, and inexpensive. The objective of the present work is to assess the performance of a newly designed, portable, low cost and network-ready ultrasonic imaging device which can be taken to rural and underserved locations for medical diagnostic and screening: SonicCare. First, a brief description of its advantages and a comparison against two commercial ultrasonography devices while scanning a tissue mimicking phantom is presented and discussed. For experimental quality assurance, the dead zone, spatial measurement calibration, spatial resolution, functional resolution and dynamic range were tested while scanning an ATS 539 multipurpose rubber based phantom. Also, clinical image quality tests were performed on two local and one rural health services facilities. Measured dimensions and image based screening on liver, gallbladder and kidney were performed. To assess the correlation between devices' measurements, the Bland-Altman method was used. The experimental quality assurance tests revealed a mean measurement error of 4.7% and a coefficient of variability of 5% while the clinical imaging test revealed that the cross-correlation precision against a clinical non-portable device was ±8.5 mm.

Pneumonia is one of the most common acute respiratory infections among pediatric populations worldwide. Ultrasound is becoming increasingly important in the diagnosis of lung diseases as a more portable and safer alternative to X-ray imaging. In the current work, we present a new automatic system for detection of B-lines, one of the distinctive features of pneumonic ultrasound scans, using amplitude modulation-frequency modulation (AM-FM). Features are evaluated on 109 videos obtained from 100 pediatric patients using a Verasonics V1 scanner. Further, the results were compared to the ones obtained with a previously published spectral feature (SF) method. Sensitivities of 92% and 83% and specificities of 91% and 70% were obtained on zone-1 and zone-2 of the lungs, respectively. In contrast, the SF method provided sensitivities of 72% and 68% in zone-1 and zone-2, respectively, and specificities of 68% and 46% in zone-1 and zone-2, respectively. In addition, the AM-FM method allowed increasing the F1-score when compared to the SF method from 70% to 87% and from 61% to 78% in zone-1 and zone-2, respectively. The results suggest the proposed method may be useful for the computer assisted diagnosis of pneumonia.

Pneumonic lung sonograms are known to include vertical comet-tail artifacts called B-lines. In this study, the potential of histogram properties from lung ultrasound images for the automatic identification of B-line artifacts is explored. Five histogram features (skewness, kurtosis, standard deviation, energy and average) were calculated for intercostal spaces. The sample consisted of 15 positive- and 15 negative-diagnosed B-mode videos selected by a medical expert and captured in a local pediatric health institute. For each frame, an initial domain of interest (DOI) starting from the pleural line is automatically outlined. The pleura is detected by a brightness based thresholding. Smaller regions containing the intercostal spaces inside the DOI are then outlined and histogram features are estimated. The potential classification of properties was evaluated independently, in pairs and using the group of 5. For single feature analysis, the optimal threshold was selected based on ROC (receiver operator characteristic) curve. For studying features in pairs a support vector machine (SVM) analysis using a RBF kernel was performed. Finally, for studying the five features, PCA (principal component analysis) was useful to determine the two principal components and apply an algorithm able to identify a B-line in the intercostal space. The results revealed that energy performed best as discriminator when using a single feature with 77% sensitivity, 75% specificity and 75% accuracy. When using features in pairs, average and skewness performed best with 93% sensitivity, 86% specificity and 88% accuracy. Finally, analyzing the 5 features, the results were 100% sensitivity, 98% specificity and 98% accuracy.

Cavity volume is an important clinical index for the assessment of the healing process and effectiveness of treatment applied on chronic ulcers. Recently, 3D scanners have proven to effectively track ulcer’s volume evolution. However, photogrammetry presents itself as a low cost and portable alternative. We conducted an inter-laboratory comparative study between photogrammetric and 3D scanner-based volume estimation of small skin ulcers. A total of 20 Cutaneous Leishmaniasis ulcers’ virtual models were generated using a commercial laser scanner and a full-HD portable camera. The reconstruction from videos was performed using comercial and open-source software (i.e., Agisoft Photoscan and VisualSFM). The results revealed similar performance with a median deviation of 16.18% and 21.10% (compared to 3DScan-based volume estimation) using VisualSFM and PhotoScan respectively. In addition, both methods proved similar efficiency in the assessment of healing ulcers when compared to 3D-scanner.

Pediatric Pneumonia is one of the principal causes of death by year on children under the age of five worldwide. The diagnosis is commonly made by clinical criteria with support from imaging tools like radiography. Lung ultrasound has been considered a low-cost and portable alternative for pneumonia imaging; however, interpretation is subjective and requires adequate training. In the present work, a pneumonia detection algorithm based on the measurement of the fundamental bandwidth downshift over depth of ultrasound radiofrequency (RF) signals is presented. RF-data was obtained from lung ultrasound samples of children aged between six months and five years. Sampling was performed using a 6.6 MHz linear transducer. The sample consisted of 10 positive-and 10 negative-diagnosed RF cine-loops selected by a medical expert and captured in a local pediatric health institute. For each frame, several regions of interest were outlined starting from the pleural line. Corresponding functions for each RF-line of the maximum frequency decrement rate over depth from the fundamental spectra at a fixed bandwidth were estimated and linearly fitted. Finally, a descriptor function was build concatenating all fitted values from the RF-lines for each frame respectively. Each descriptor function was later thresholded to differentiate between healthy and pneumonic regions frame-wise. The optimal threshold was found to be 0.46 MHz/cm and was selected based on a receiver operator characteristic (ROC) curve analysis. The results revealed an accuracy rate higher than 90% on the sample.

According to World Health Organization, pneumonia is the respiratory disease with the highest pediatric mortality rate accounting for 15% of all deaths of children under 5 years old worldwide. The diagnosis of pneumonia is commonly made by clinical criteria with support from ancillary studies and also laboratory findings. Chest imaging is commonly done with chest X-rays and occasionally with a chest CT scan. Lung ultrasound is a promising alternative for chest imaging; however, interpretation is subjective and requires adequate training. In the present work, a two-class classification algorithm based on four Gray-level co-occurrence matrix texture features (i.e., Contrast, Correlation, Energy and Homogeneity) extracted from lung ultrasound images from children aged between six months and five years is presented. Ultrasound data was collected using a L14-5/38 linear transducer. The data consisted of 22 positive- and 68 negative-diagnosed B-mode cine-loops selected by a medical expert and captured in the facilities of the Instituto Nacional de Salud del Nino (Lima, Peru), for a total number of 90 videos obtained from twelve children diagnosed with pneumonia. The classification capacity of each feature was explored independently and the optimal threshold was selected by a receiver operator characteristic (ROC) curve analysis. In addition, a principal component analysis was performed to evaluate the combined performance of all the features. Contrast and correlation resulted the two more significant features. The classification performance of these two features by principal components was evaluated. The results revealed 82% sensitivity, 76% specificity, 78% accuracy and 0.85 area under the ROC.

The quantitative imaging of attenuation coefficients slope (ACS) has the potential to improve medical diagnostics. However, attempts to characterize ACS using pulse-echo data have been limited by the large statistical variations in the estimates. Previous studies demonstrated that it is possible to extend the trade-off between variance and spatial resolution of quantitative ultrasound, spectral-based parameters by the use of full angular (i.e., 360°) spatial compounding (FASC). In the present work, the use of FASC has been extended to the estimation of ACS and its performance has been experimentally evaluated using two physical phantoms. The ACSs of the background and inclusion regions were estimated using insertion loss measurements to be 0.41 and 0.75 dB/cm/MHz for Phantom #1, and 0.54 and 1.04 dB/cm/MHz for Phantom #2, respectively. Pulse-echo data were collected using a 7.5 MHz, f/4 transducer at 30 angles of view uniformly distributed between 0 and 360°. Single view ACS maps were generated using a spectral log difference method with 0.6 by 0.6 mm data blocks. The FASC images were constructed by assigning to a pixel the median of its corresponding estimates from all 30 angles of view. The reduction in the variance of the FASC estimates compared to the variance of estimates from a single view (i.e., variance averaged from the 30 single views) in the inclusion and background regions were 89.18% and 88.71% for Phantom #1 and 92.33% and 86.98% for Phantom #2. Moreover, in all the cases the estimation bias in the inclusion and background regions using FASC was lower than 9.0%. These results suggest that the variance of attenuation coefficient slope estimation can be significantly reduced without sacrificing spatial resolution by the use of full angular spatial compounding.

The incidence rate of diagnosed thyroid cancer has increased over the last decades. Although ultrasonic imaging has increased the malignancy detection rate, current ultrasonography markers do not provide a sufficient level of diagnostic accuracy to replace the fine needle aspiration biopsy. Recently, studies have reported that significant differences were observed in the values of quantitative ultrasound (QUS) parameters derived from a thyroid cancer rodent model between normal/benign and malignant tissues. In the present study, the performance of a multi-parametric classification for the differentiation of thyroid cancer in this rodent model has been evaluated. The experimental database consisted of 32 mice having different predispositions to developing thyroid abnormalities; 6 of them developed thyroid cancer papillary carcinoma (PTC), 5 follicular variant papillary thyroid carcinoma (FV-PTC), 6 developed benign tumors (c-cell adenoma) and 15 did not develop any thyroid abnormalities. Backscattered data was obtained from excised thyroid tissues using a 40 MHz, f/3 single element transducer. A total of five QUS parameters were derived from the ultrasound data: two from backscatter coefficients (i.e., the effective scatterer diameter (ESD) and effective acoustic concentration (EAC)), two from envelope statistics (i.e., the μ and k parameters), and one from ultrasound attenuation (i.e., attenuation coefficient slope). A two-class classification between normal/benign and malignant cases was performed using linear discriminant analysis with both one- and two-dimensional feature spaces. When using a two-dimensional feature space, it was found that the combination of EAC and 10/μ resulted in both a sensitivity and specificity of 100%.

The incidence of diagnosed thyroid cancer has increased significantly over the last decades. Although advances in ultrasonic imaging have increased the malignancy detection rate, current ultrasonic imaging markers do not provide a sufficient level of diagnostic accuracy to replace biopsy. Recent studies suggest that ultrasound parameters derived from backscatter coefficients may allow differentiating among different types of thyroid tumors and normal tissues in a rodent model ex vivo. In this work, the potential use of attenuation coefficient (AC) estimates for the same purpose was explored. A sample set of 24 excised mice thyroids were scanned using a 40-MHz, f/3 single element transducer. The experimental dataset contained six animals that developed papillary thyroid carcinoma (PTC), five that developed follicular variant papillary thyroid carcinoma (FV-PTC), five that developed c-Cell adenoma (c-Cell) and eight that did not develop thyroid abnormalities (control). AC slope maps were generated with a spectral log difference method using 0.5mm by 0.5mm data blocks. Outliers of each slice due to artifacts in AC estimation were discarded using the Thompson Tau method. Finally, a Kruskal-Wallis test was conducted to analyze if statistically significant differences in the mean AC slope among the four groups existed. The median and interquartile range for each group were 1.29 and 0.22 dB/cm-MHz for the control group, 1.64 and 0.09 dB/cm-MHz for c-Cell, 1.16 and 0.12 dB/cm-MHz for PTC and 1.33 and 0.08 dB/cm-MHz for FV-PTC, respectively. These values are consistent with previous reports of attenuation in thyroid tissues. The Kruskal-Wallis test reported statistically significant differences between the c-Cell group and the other groups of study and between the PTC and FV-PTC groups (p