Image registration and atlas-based segmentation of cardiac outflow velocity profiles

Kalinić, Hrvoje; Lončarić, Sven; Čikeš, Maja; Miličić, Davor; Bijnens, Bart

doi:10.1016/j.cmpb.2010.11.001

Cited by 8 publications

(6 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Gaillard et al (2010) describe the aortic and mitral Doppler wave segmentation using a more advanced active contour method, initialized with a shape of the centers of divergence of the gradient vector flow field, reporting sensitivity of the method to the image contrast. Motivated by the clinicians' need for not only fast and automatic delineation but also for automatic extraction of condition-descriptive mathematically-derived features from the aortic valve velocity profile shape, Kalinić et al (2009aKalinić et al ( ,b, 2012 applied and developed several methods for atlas-based segmentation of Doppler velocity images. Some of those are adapted and incorporated here as a part of our novel hybrid method.…”

Section: Segmentation Of Aortic Doppler Velocity Profilesmentioning

confidence: 99%

“…Using an apical 5-chamber view, the scanner records the realtime blood velocity through the aortic valve during several heart cycles and stores the data digitally in "raw" Dicom format. Further analysis of the images can be performed with an EchoPAC workstation (GE Healthcare) and relevant information can be exported for further analysis (Kalinić et al (2012)).…”

Section: Clinically Obtained Aortic Outflow Doppler Imagesmentioning

confidence: 99%

See 1 more Smart Citation

A computational model-based approach for atlas construction of aortic Doppler velocity profiles for segmentation purposes

Balicevic

Kalinić

Lončarić

et al. 2018

Biomedical Signal Processing and Control

Self Cite

View full text Add to dashboard Cite

Echocardiography is the leading imaging modality for cardiac disorders in clinical practice. During an echocardiographic exam, geometry and blood flow are quantified in order to assess cardiac function. In clinical practice, these imagebased measurements are currently performed manually. An automated approach is needed if more advanced analysis is desired. In this article, we propose a new hybrid framework for the construction of a disease-specific atlas to improve Doppler aortic outflow velocity profile segmentation. The proposed method is based on combining realistic computational simulations of the cardiovascular system for common cardiac conditions (using Cir-cAdapt) with a validated image-based atlas construction method. The coupling is realized via model-based generation of echocardiographic images of virtual populations with a statistically approved parameter variation. We created virtual populations of 100 healthy individuals and 100 patients with aortic stenosis, synthesized their aortic Doppler velocity images and constructed the corresponding atlases. We validated atlases' performances by comparing their segmentation of real clinical images with the manually segmented ground truth. The experimental results show that the segmentation accuracy obtained using the proposed atlases is comparable to the accuracy obtained using classical clinical image-based atlases. Moreover, this framework eliminates the time-consuming acquisition of a sufficient number of representative images in clinical practice, offering a substantial time efficiency and flexibility in creating a disease specific atlas and ensuring an observer-independent automated segmentation. The proposed approach can easily be extended towards the creation of atlases for segmenting any Doppler trace in the cardiovascular circulation in a specific disease.

show abstract

Section: Segmentation Of Aortic Doppler Velocity Profilesmentioning

confidence: 99%

Section: Clinically Obtained Aortic Outflow Doppler Imagesmentioning

confidence: 99%

A computational model-based approach for atlas construction of aortic Doppler velocity profiles for segmentation purposes

Balicevic

Kalinić

Lončarić

et al. 2018

Biomedical Signal Processing and Control

Self Cite

View full text Add to dashboard Cite

show abstract

“…Medical image auto‐contouring has been an active field of study since almost as long as tomographic imaging has been in use 5 . Commonly used algorithm frameworks include adaptive level sets 6 active contours, 7 and atlas‐based contouring 8 . Commercially‐available tools using atlas‐based contouring have been evaluated and have remained in clinical use 9 .…”

Section: Introductionmentioning

confidence: 99%

Human factors in the clinical implementation of deep learning‐based automated contouring of pelvic organs at risk for MRI‐guided radiotherapy

Abdulkadir,

Luximon,

Morris

et al. 2023

Medical Physics

View full text Add to dashboard Cite

PurposeDeep neural nets have revolutionized the science of auto‐segmentation and present great promise for treatment planning automation. However, little data exists regarding clinical implementation and human factors. We evaluated the performance and clinical implementation of a novel deep learning‐based auto‐contouring workflow for 0.35T magnetic resonance imaging (MRI)‐guided pelvic radiotherapy, focusing on automation bias and objective measures of workflow savings.MethodsAn auto‐contouring model was developed using a UNet‐derived architecture for the femoral heads, bladder, and rectum in 0.35T MR images. Training data was taken from 75 patients treated with MRI‐guided radiotherapy at our institution. The model was tested against 20 retrospective cases outside the training set, and subsequently was clinically implemented. Usability was evaluated on the first 30 clinical cases by computing Dice coefficient (DSC), Hausdorff distance (HD), and the fraction of slices that were used un‐modified by planners. Final contours were retrospectively reviewed by an experienced planner and clinical significance of deviations was graded as negligible, low, moderate, and high probability of leading to actionable dosimetric variations. In order to assess whether the use of auto‐contouring led to final contours more or less in agreement with an objective standard, 10 pre‐treatment and 10 post‐treatment blinded cases were re‐contoured from scratch by three expert planners to get expert consensus contours (EC). EC was compared to clinically used (CU) contours using DSC. Student's t‐test and Levene's statistic were used to test statistical significance of differences in mean and standard deviation, respectively. Finally, the dosimetric significance of the contour differences were assessed by comparing the difference in bladder and rectum maximum point doses between EC and CU before and after the introduction of automation.ResultsMedian (interquartile range) DSC for the retrospective test data were 0.92(0.02), 0.92(0.06), 0.93(0.06), 0.87(0.04) for the post‐processed contours for the right and left femoral heads, bladder, and rectum, respectively. Post‐implementation median DSC were 1.0(0.0), 1.0(0.0), 0.98(0.04), and 0.98(0.06), respectively. For each organ, 96.2, 95.4, 59.5, and 68.21 percent of slices were used unmodified by the planner. DSC between EC and pre‐implementation CU contours were 0.91(0.05*), 0.91*(0.05*), 0.95(0.04), and 0.88(0.04) for right and left femoral heads, bladder, and rectum, respectively. The corresponding DSC for post‐implementation CU contours were 0.93(0.02*), 0.93*(0.01*), 0.96(0.01), and 0.85(0.02) (asterisks indicate statistically significant difference). In a retrospective review of contours used for planning, a total of four deviating slices in two patients were graded as low potential clinical significance. No deviations were graded as moderate or high. Mean differences between EC and CU rectum max‐doses were 0.1 ± 2.6 Gy and −0.9 ± 2.5 Gy for pre‐ and post‐implementation, respectively. Mean differences between EC and CU bladder/bladder wall max‐doses were −0.9 ± 4.1 Gy and 0.0 ± 0.6 Gy for pre‐ and post‐implementation, respectively. These differences were not statistically significant according to Student's t‐test.ConclusionWe have presented an analysis of the clinical implementation of a novel auto‐contouring workflow. Substantial workflow savings were obtained. The introduction of auto‐contouring into the clinical workflow changed the contouring behavior of planners. Automation bias was observed, but it had little deleterious effect on treatment planning.

show abstract

“…Several studies have targeted automated methods for spectral envelope segmentation and interpretation of Doppler images using signal processing and machine learning approaches. Such as low-level image-processing based methods [35,42], texture filter analysis [219] and thresholding and edge detection [43][44][45][46][47][48], contour-based and model-based methods for Doppler segmentation [36][37][38][39]49] and traditional machine learning [40,41].…”

Section: Related Workmentioning

confidence: 99%

“…• measuring peak velocitites from Tissue Doppler Imaging (TDI) images [35][36][37][38][39][40][41][42][43][44][45][46][47][48][49] (Chapter…”

Section: Introductionmentioning

confidence: 99%

A computer vision pipeline for fully automated echocardiogram interpretation

Lane¹

View full text Add to dashboard Cite

Cardiovascular disease is the leading cause of global mortality and continues to place a significant burden, in economic and resource terms, upon health services. A 2-dimensional transthoracic echocardiogram captures high spatial and temporal images and videos of the heart and is the modality of choice for the rapid assessment of heart function and structure due to it’s non-invasive nature and lack of ionising radiation. The challenging process of analysing echocardiographic images is currently manually performed by trained experts, though this process is vulnerable to intra- and inter-observer variability and is highly time-consuming. Additionally, echocardiographic images suffer from varying degrees of noise and vary drastically in terms of image quality. Exponential advancements in the fields of artificial intelligence, deep learning and computer vision have enabled the rapid development of automated systems capable of high-precision tasks, often out-performing human experts. This thesis aims to investigate the applicability of applying deep learning methods to automate key processes in the modern echocardiographic laboratory. Namely, view classification, quality assessment, cardiac phase detection, segmentation of the left ventricle and keypoint detection on tissue Doppler imaging strips. State-of-the-art deep learning architectures were applied to each task, and evaluated against ground-truth annotations provided by trained experts. The datasets used throughout each Chapter are diverse and, in some cases, have been made public for the benefit of the research community. To encourage transparency and openness, all code and model weights have been published. Should automated deep learning systems, both online (in terms of providing real-time feedback) and offline (behind the scenes), become integrated within clinical practice, there is great potential for improved accuracy and efficiency, thus improving patient outcomes. Furthermore, health services could save valuable resources such as time and money.

show abstract

Image registration and atlas-based segmentation of cardiac outflow velocity profiles

Cited by 8 publications

References 35 publications

A computational model-based approach for atlas construction of aortic Doppler velocity profiles for segmentation purposes

A computational model-based approach for atlas construction of aortic Doppler velocity profiles for segmentation purposes

Human factors in the clinical implementation of deep learning‐based automated contouring of pelvic organs at risk for MRI‐guided radiotherapy

A computer vision pipeline for fully automated echocardiogram interpretation

Contact Info

Product

Resources

About