Purpose
Automated delineation of structures and organs is a key step in medical imaging. However, due to the large number and diversity of structures and the large variety of segmentation algorithms, ...a consensus is lacking as to which automated segmentation method works best for certain applications. Segmentation challenges are a good approach for unbiased evaluation and comparison of segmentation algorithms.
Methods
In this work, we describe and present the results of the Head and Neck Auto‐Segmentation Challenge 2015, a satellite event at the Medical Image Computing and Computer Assisted Interventions (MICCAI) 2015 conference. Six teams participated in a challenge to segment nine structures in the head and neck region of CT images: brainstem, mandible, chiasm, bilateral optic nerves, bilateral parotid glands, and bilateral submandibular glands.
Results
This paper presents the quantitative results of this challenge using multiple established error metrics and a well‐defined ranking system. The strengths and weaknesses of the different auto‐segmentation approaches are analyzed and discussed.
Conclusions
The Head and Neck Auto‐Segmentation Challenge 2015 was a good opportunity to assess the current state‐of‐the‐art in segmentation of organs at risk for radiotherapy treatment. Participating teams had the possibility to compare their approaches to other methods under unbiased and standardized circumstances. The results demonstrate a clear tendency toward more general purpose and fewer structure‐specific segmentation algorithms.
Full text
Available for:
BFBNIB, FZAB, GIS, IJS, KILJ, NLZOH, NUK, OILJ, SAZU, SBCE, SBMB, UL, UM, UPUK
Objectives
To evaluate the predictive value of volumetric bone mineral density (BMD) assessment of the lumbar spine derived from phantomless dual-energy CT (DECT)-based volumetric material ...decomposition as an indicator for the 2-year occurrence risk of osteoporosis-associated fractures.
Methods
L1 of 92 patients (46 men, 46 women; mean age, 64 years, range, 19–103 years) who had undergone third-generation dual-source DECT between 01/2016 and 12/2018 was retrospectively analyzed. For phantomless BMD assessment, dedicated DECT postprocessing software using material decomposition was applied. Digital files of all patients were sighted for 2 years following DECT to obtain the incidence of osteoporotic fractures. Receiver operating characteristic (ROC) analysis was used to calculate cut-off values and logistic regression models were used to determine associations of BMD, sex, and age with the occurrence of osteoporotic fractures.
Results
A DECT-derived BMD cut-off of 93.70 mg/cm
3
yielded 85.45% sensitivity and 89.19% specificity for the prediction to sustain one or more osteoporosis-associated fractures within 2 years after BMD measurement. DECT-derived BMD was significantly associated with the occurrence of new fractures (odds ratio of 0.8710, 95% CI, 0.091–0.9375,
p
< .001), indicating a protective effect of increased DECT-derived BMD values. Overall AUC was 0.9373 (CI, 0.867–0.977,
p
< .001) for the differentiation of patients who sustained osteoporosis-associated fractures within 2 years of BMD assessment.
Conclusions
Retrospective DECT-based volumetric BMD assessment can accurately predict the 2-year risk to sustain an osteoporosis-associated fracture in at-risk patients without requiring a calibration phantom. Lower DECT-based BMD values are strongly associated with an increased risk to sustain fragility fractures.
Key Points
•Dual-energy CT–derived assessment of bone mineral density can identify patients at risk to sustain osteoporosis-associated fractures with a sensitivity of 85.45% and a specificity of 89.19%.
•The DECT-derived BMD threshold for identification of at-risk patients lies above the American College of Radiology (ACR) QCT guidelines for the identification of osteoporosis (93.70 mg/cm
3
vs 80 mg/cm
3
).
Full text
Available for:
EMUNI, FIS, FZAB, GEOZS, GIS, IJS, IMTLJ, KILJ, KISLJ, MFDPS, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, SBMB, SBNM, UKNU, UL, UM, UPUK, VKSCE, VSZLJ, ZAGLJ
Purpose
The acquisition conditions of medical imaging are often precisely defined, leading to a high homogeneity among different data sets. Nonetheless, outliers or artefacts still appear and need to ...be reliably detected to ensure a reliable diagnosis. Thus, the algorithms need to handle small sample sizes especially, when working with domain specific imaging modalities.
Methods
In this work, we suggest a pipeline for the detection and segmentation of light pollution in near-infrared fluorescence optical imaging (NIR-FOI), based on a small sample size. NIR-FOI produces spatio-temporal data with two spatial and one temporal dimension. To calculate a two-dimensional light pollution map for the entire image stack, we combine
region growing
and
k-nearest neighbours
(kNN), which classifies pixels into fore- and background by its entire temporal component. Thus, decision-making on reduced data is omitted.
Results
We achieved a
F
1
score of 0.99 for classifying a data set as light polluted or pollution-free. Additionally, we reached a total
F
1
score of 0.90 for detecting regions of interest within the polluted data sets. Finally, an average Dice’s coefficient measuring the segmentation performance over all polluted data sets of 0.80 was accomplished.
Conclusions
A Dice’s coefficient of 0.80 for the area segmentation does not seem perfect. However, there are two main factors, besides true prediction errors, lowering the score: Segmentation mistakes on small areas lead to a rapid decrease in the score and labelling errors due to complex data. However, in combination with the light-polluted data set and pollution area detection, these results can be considered successful and play a key role in our general goal: Exploiting NIR-FOI for the early detection of arthritis within hand joints.
Full text
Available for:
EMUNI, FIS, FZAB, GEOZS, GIS, IJS, IMTLJ, KILJ, KISLJ, MFDPS, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, SBMB, SBNM, UKNU, UL, UM, UPUK, VKSCE, ZAGLJ
In this work, we investigate the possibility of automatically and objectively evaluate an initialized therapy for psoriatic arthritis patients (therapy monitoring). Near infrared fluorescence optical ...imaging (NIR-FOI) is used to visualise the microcirculation in the hands, giving insights into possible inflammations (e.g. synovitis). Due to the time consuming and expansive data acquisition, we firstly investigate the feasibility of NIR-FOI in this scope of application using retrospective data. To increase the data quality and harmonize the heterogeneous patient population, certain inclusion criteria must be met, leaving only a limited amount of data sets for data analysis. A neural network was trained to segment the hands into areas of interest (e.g. joints and fingers) and an image registration pipeline was developed to compare identified regions of interest among different patient visits. It is shown that data from different visits are not directly comparable and thus, needs to be normalised. Due to the limited amount of meta data, only a simple normalisation can be performed. Even though with this normalisation the therapy monitoring cannot be carried out, the normalised data shows an increased agreement to a medical label in comparison to unnormalised data. Therefore, we conclude that further research can lead to a robust therapy monitoring algorithm.
Background
Dual-source dual-energy computed tomography (DECT) offers the potential for opportunistic osteoporosis screening by enabling phantomless bone mineral density (BMD) quantification. This ...study sought to assess the accuracy and precision of volumetric BMD measurement using dual-source DECT in comparison to quantitative CT (QCT).
Methods
A validated spine phantom consisting of three lumbar vertebra equivalents with 50 (L1), 100 (L2), and 200 mg/cm
3
(L3) calcium hydroxyapatite (HA) concentrations was scanned employing third-generation dual-source DECT and QCT. While BMD assessment based on QCT required an additional standardised bone density calibration phantom, the DECT technique operated by using a dedicated postprocessing software based on material decomposition without requiring calibration phantoms. Accuracy and precision of both modalities were compared by calculating measurement errors. In addition, correlation and agreement analyses were performed using Pearson correlation, linear regression, and Bland-Altman plots.
Results
DECT-derived BMD values differed significantly from those obtained by QCT (
p
< 0.001) and were found to be closer to true HA concentrations. Relative measurement errors were significantly smaller for DECT in comparison to QCT (L1, 0.94%
versus
9.68%; L2, 0.28%
versus
5.74%; L3, 0.24%
versus
3.67%, respectively). DECT demonstrated better BMD measurement repeatability compared to QCT (coefficient of variance < 4.29% for DECT, < 6.74% for QCT). Both methods correlated well to each other (
r
= 0.9993; 95% confidence interval 0.9984–0.9997;
p
< 0.001) and revealed substantial agreement in Bland-Altman plots.
Conclusions
Phantomless dual-source DECT-based BMD assessment of lumbar vertebra equivalents using material decomposition showed higher diagnostic accuracy compared to QCT.
Cephalometric analysis is an important method in orthodontics for the diagnosis and treatment of patients. It is performed manually in clinical practice, therefore automation of this time consuming ...task would be of great assistance. In order to provide dentists with such tools, a robust and accurate identification of the necessary landmarks is required. However, poor image quality of lateral cephalograms like low contrast or noise make this task difficult. In this paper, an approach for automatic landmark localization is presented and used to find 19 landmarks in lateral cephalometric images. An initial predicting of the individual landmark locations is done by using a 2-D coupled shape model to utilize the spatial relation between landmarks and other anatomical structures. These predictions are refined with a Hough Forest to determine the final landmark location. The approach achieves competitive performance with a successful detection rate of 70.24% on 250 images for the clinically relevant 2mm accuracy range.
For segmenting complex structures like vertebrae, a priori knowledge by means of statistical shape models (SSMs) is often incorporated. One of the main challenges using SSMs is the solution of the ...correspondence problem. In this work we present a generic automated approach for solving the correspondence problem for vertebrae. We determine two closed loops on a reference shape and propagate them consistently to the remaining shapes of the training set. Then every shape is cut along these loops and parameterized to a rectangle. There, we optimize a novel combined energy to establish the correspondences and to reduce the unavoidable area and angle distortion. Finally, we present an adaptive resampling method to achieve a good shape representation. A qualitative and quantitative evaluation shows that using our method we can generate SSMs of higher quality than the ICP approach.
Full text
Available for:
FIS, FZAB, GEOZS, GIS, IJS, IMTLJ, KILJ, KISLJ, MFDPS, NUK, OBVAL, OILJ, PNG, SAZU, SBCE, SBJE, SBMB, SBNM, UKNU, UL, UM, UPUK, VKSCE, ZAGLJ
Proper treatment of prostate cancer is essential to increase the survival chance. In this sense, numerous studies show how important the communication between all stakeholders in the clinic is. This ...communication is difficult because of the lack of conventions while referring to the location where a biopsy for diagnosis was taken. This becomes even more challenging taking into account that experts of different fields work on the data and have different requirements. In this paper a web-based communication tool is proposed that incorporates a visualization of the prostate divided into 27 segments according to the PI-RADS protocol. The tool provides 2 working modes that consider the requirements of radiologist and pathologist while keeping it consistent. The tool comprises all relevant information given by pathologists and radiologists, such as, severity grades of the disease or tumor length. Everything is visualized using a colour code for better undestanding.
After a liver tumor intervention the medical doctor has to compare both pre and postoperative CT acquisitions to ensure that all carcinogenic cells are destroyed. A correct assessment of the ...intervention is of vital importance, since it will reduce the probability of tumor recurrence. Some methods have been proposed to support the medical doctors during the assessment process, however, all of them focus on secondary tumors. In this paper a tool is presented that enables the outcome validation for both primary and secondary tumors. Therefore, a multiphase registration (preoperative arterial and portal phases) followed by a registration between the pre and postoperative CT images is carried out. The first registration is in charge of the primary tumors that are only visible in the arterial phase. The secondary tumors will be incorporated in the second registration step. Finally, the part of the tumor that was not covered by the necrosis is quantified and visualized. The method has been tested in 9 patients, with an average registration error of 1.41 mm.