Comparison of Otsu and an adapted Chan–Vese method to determine thyroid active volume using Monte Carlo generated SPECT images

본 논문에서는 주어진 영상의 그레이 레벨에 대한 통계적 정보와 구간값 퍼지집합에 기반을 둔 새로운 임계값 선택 방법을 제안한다. 제안한 임계값 선택 방법에서 구간값 퍼지집합은 영상의 픽셀과 그들이 속하는 영역, 즉 물체와 배경 간의 관계를 더욱 명확하게 나타내기 위해서 사용되고, 통계적 정보는 구간값 퍼지집합의 규칙과 파티션을 결정하기 위해서 이용된다. 제안한 방법의 타당성을 보이기 위해 다양한 형태의 히스토그램을 가진 5개의 테스트 영상들을 기존의 임계값 선택방법인 Otsu 방법과 Huang과 Wang의 방법과 비교하였다. In this paper, we propose a novel threshold selection method based on statistical information on gray-levels of given images and interval-valued fuzzy sets. In the proposed threshold selection method, the interval-valued fuzzy set is used to represent more definitely the relationship between a pixel and its belonging region, that is, the object and the background. Also the statistical information on gray-level is used to determine the rules and partitions of interval-valued fuzzy sets. To show the validity of the proposed method, we compared the performance of the proposed with those of conventional methods such as Otsu's method, Huang and Wang's method applied to 5 test images with various types of histograms.

Research Highlights: This paper proposes a new method for hemispherical forest canopy image segmentation. The method is based on a deep learning methodology and provides a robust and fully automatic technique for the segmentation of forest canopy hemispherical photography (CHP) and gap fraction (GF) calculation. Background and Objectives: CHP is widely used to estimate structural forest variables. The GF is the most important parameter for calculating the leaf area index (LAI), and its calculation requires the binary segmentation result of the CHP. Materials and Methods: Our method consists of three modules, namely, northing correction, valid region extraction, and hemispherical image segmentation. In these steps, a core procedure is hemispherical canopy image segmentation based on the U-Net convolutional neural network. Our method is compared with traditional threshold methods (e.g., the Otsu and Ridler methods), a fuzzy clustering method (FCM), commercial professional software (WinSCANOPY), and the Habitat-Net network method. Results: The experimental results show that the method presented here achieves a Dice similarity coefficient (DSC) of 89.20% and an accuracy of 98.73%. Conclusions: The method presented here outperforms the Habitat-Net and WinSCANOPY methods, along with the FCM, and it is significantly better than the Otsu and Ridler threshold methods. The method takes the original canopy hemisphere image first and then automatically executes the three modules in sequence, and finally outputs the binary segmentation map. The method presented here is a pipelined, end-to-end method.

Cancer is a leading cause of death among men and women nowadays all over the world. Breast cancer is a most common form of cancer originated from breast tissue among women. Most frequent type of breast cancer is ductal carcinoma in situ (DCIS) and most frequent symptoms of DCIS recognized by mammography are clusters of Microcalcifications (MCCs). Automatic detection of Microcalcifications is an important task to prevent and treat the disease. In this paper, an effective approach for automatic detection of Microcalcifications in digitized mammograms is proposed. The proposed approach is based on fuzzy 2-partition entropy and mathematical morphology. In the proposed approach, first phase uses fuzzy Gaussian membership function for mammogram fuzzification. In this phase, fuzzy 2-partition entropy approach is used to find bandwidth of the Gaussian function. After this, mathematical morphological enhancement approach is used to enhance the contrast of Microcalcifications in mammograms. Finally, Microcalcifications are located using Otsu threshold selection method. Experiments have been conducted on images of mini-MIAS database (Mammogram Image Analysis Society database (UK)). In order to validate the results, several different kinds of standard test images (fatty, fatty-glandular and denseglandular) of mini-MIAS database are considered. Experimental results demonstrate that the proposed approach has an ability to detect Microcalcifications even in dense mammograms. The results of proposed approach are quite promising. The proposed approach can be a part of developing a computer aided decision (CAD) system for early detection of breast cancer.

Maximum permitted levels of radiocesium in marketed animal feedstuffs have been specified by the European Commission for use in the event of a future nuclear accident. However, more specific guidance is required for use under UK conditions. Using information on typical animal diets, relative activity concentrations in feedstuffs and feed-to-product transfer, practical working levels have been derived for (134,137)Cs and 90Sr in animal feedstuffs and drinking water; application of these levels should ensure that activity concentrations in milk, meat, or eggs do not exceed the relevant Council Food Intervention Level. Despite the complexity of some animal diets, only one or two feedstuffs typically contribute 5% or more to intakes of activity. The working levels for radiocesium in these feedstuffs ranged from a factor of 20 higher than the maximum permitted levels to a factor of around 20 lower. In most cases, the maximum permitted levels are unnecessarily cautious. Working levels for 90Sr in feedstuffs are generally higher than those for radiocesium, except for feedstuffs for dairy cattle and laying hens. Factors affecting the practicability and effectiveness of changes in diet as a countermeasure to reduce activity concentrations in animal products were also considered within this study.

Evaluation and follow up of small pulmonary nodules has become commonplace with advancing PET and CT technology. Limits in system resolution, recovery coefficient estimations and transaxial ROI determinations often make obtaining useful SUV data difficult. We demonstrate a method which may be more accurate for making these determinations. It is assumed that the nodule can be distinguished clearly on CT and that it resides in a region of homogenous background PET uptake. By integrating the PET activity over a large region including the small nodule and determining the volume of the lesion (V/sub lesion/) from thin slice CT images the activity concentration in the lesion (AC/sub lesion/) can be determined if the background activity concentration (AC/sub background/) is determined away from the lesion. The total integrated activity (IA) in the ROI is: IA=AC/sub lesion//spl middot/V/sub lesion/+AC/sub background/$ 8SV/sub background/. Alternatively the ratio of the activity concentration in the lesion to the background is: [(IA/AC/sub background/)-V/sub background/]/V/sub l /e/sub sion/. Results for activity ratios using a hollow spheres phantom indicate this method gives the correct value for the smallest sphere (activity volume=0.6cc) whereas the ratio using usual SUV determinations is spuriously low by a factor of up to /spl sim/2. Clinical results for a 1cc volume SPN indicate a tumor to background ratio of /spl sim/20 compared with a ratio of 15 using the maximum SUV, similar to the behavior seen in phantom measurements.

<strong class="journal-contentHeaderColor">Abstract.</strong> We observed ³H activity concentrations and the ¹³⁷Cs activity concentrations during the SoSo 5 rivers cruise in 2014 and at the Tomioka port in 2014&ndash;2018. The ³H activity concentrations at coastal stations located close to the Fukushima coast ranged from 90&thinsp;Bq&thinsp;m^&minus;3 to 175&thinsp;Bq&thinsp;m^&minus;3, and decreased between 67&thinsp;Bq&thinsp;m^&minus;3 to 83&thinsp;Bq&thinsp;m^&minus;3 at the stations located 12&ndash;16&thinsp;km from the coast. The ³H activity concentration at the estuaries and ports, except at 56 north canal of the FNPP1 site, are around 200&ndash;500&thinsp;Bq&thinsp;m^&minus;3 and slightly lower than the ³H activity concentration of 500&ndash;600&thinsp;Bq&thinsp;m^&minus;3 observed in the rivers. These gradients of the ³H activity concentrations in the coastal region might indicate the large effect of ³H flux through the rivers. On the other hand, the ³H activity concentration at 56N of the FNPP1 site was significantly high compared to the ³H activity concentration in surrounding waters both north and south of the FNPP1 site and in river waters. It should also be noted that the ³H activity concentrations were similar at the stations located both north and south of the FNPP1 site, while the ¹³⁷Cs activity concentrations were lower at the stations north of the FNPP1 site and higher at the stations south of the FNPP1 site. This indicated that the major sources of ¹³⁷Cs could be the FNPP1 site as the point source while the source of ³H should be more diffuse and linked to riverine inputs located north and south of the FNPP1 site. The ³H&thinsp;/&thinsp;¹³⁷Cs activity ratios in coastal waters were 1.2&ndash;2.2 as obtained via the slopes by standardised major axis regressions between the ³H activity concentration and the ¹³⁷Cs activity concentration of SoSo samples and Tomioka Port observed in 2014&ndash;2018, which is significantly high compared to that of the released radionuclides derived from the FNPP1 site, which was 0.01 in 2011 just after the accident. The open-water ³H activity concentration contribution to coastal waters was estimated to be 67&thinsp;±&thinsp;20&thinsp;Bq&thinsp;m^&minus;3 and 66&thinsp;±&thinsp;17&thinsp;Bq&thinsp;m^&minus;3 as the intercepts also by standardised major axis regressions. These estimates are consistent with 50&thinsp;Bq&thinsp;m^&minus;3 obtained at the Kuroshio region as the background levels of ³H activity concentration in open water. The ³H and ¹³⁷Cs fluxes to the coastal region of Fukushima based on the open-water movement, freshwater flux from the rivers based on their respective catchment, and mean monthly precipitation were estimated. The largest ³H flux is the open-water inflow from the north of the FNPP1 site and it reaches 52&thinsp;GBq&thinsp;day^&minus;1, while the rivers north of the FNPP1 site showed 3&ndash;5&thinsp;GBq&thinsp;day^&minus;1 fluxes. We obtained a ³H flux as 1.9&ndash;4.5&thinsp;GBq&thinsp;day^&minus;1 of ³H using the ³H activity concentration in the port, which is comparable with the fluxes obtained from the rivers located north of the FNPP1 site. While using ³H activity concentration at the 56 north canal of FNPP1, we obtained 28&ndash;86&thinsp;GBq&thinsp;day^&minus;1 fluxes, which is one order of magnitude larger than those estimated using ³H activity concentration in the FNPP1 port. One of the reasons could be the very high variability of the ³H levels at 56 north canal and in the port of FNPP1, explaining variable ³H/¹³⁷Cs activity ratio observed at 56 north canal and in the port of FNPP1. The ³H activity concentration of TFWT in the fish filets collected close to the FNPP1 site ranged from 97&thinsp;±&thinsp;11&thinsp;Bq&thinsp;m^&minus;3 to 144&thinsp;±&thinsp;11&thinsp;Bq&thinsp;m^&minus;3, which were similar to the ³H activity concentrations in the surrounding seawater, in agreement with the knowledge that the bioconcentration factor of ³H is approximately 1. In contrast, higher values were found in TOBT, which can be linked to life-history traits.

To quantify the partial volume effect in single photon emission tomography (SPECT) and planar images of Carlson phantom as well as providing an optimum region of interest (ROI) required to more accurately estimate the activity concentration for different sphere sizes. 131 I solution with the 161.16 kBq/ml concentration was uniformly filled into the different spheres of Carlson phantom (cold background condition) with the diameters of 7.3, 9.2, 11.4, 14.3, 17.9, 22.4 and 29.9 mm, and there was no background activity. In the hot background condition, the spheres were filled with the solution of 131 I with the 1276.5 kBq/ml addition to the background activity concentration of 161.16 kBq/ml in all the phantoms. The spheres were mounted inside the phantom and underwent SPECT and planar images. ROI was drawn closely on the boundary of each sphere image and it was extended to extract the true count. In the cold background condition, the recovery coefficient (RC) value for SPECT images ranged between 0.8 and 1.03. However, in planar imaging, the RC value was 0.72 for the smallest sphere size and it increased for larger spheres until 0.98 for 29.9 mm. In the hot background condition, the RC value for sphere diameters larger than 20 mm was overestimated more than in the cold background condition. The ROI/size required to more accurately determine activity concentration for the cold background ranged from 1.18 to 2.7. However, in the hot background condition, this ratio varied from 1.34 to 4.05. In the quantification of partial volume effects, the spill-out effect seems to play a crucial role in the distribution of the image counts beyond the boundaries of the image pixels. However, more investigations are needed to accurately characterize limitations regarding the object size, background levels, and other factors.

In this paper, we proposed and validated a novel and accurate pipeline for automatically segmenting flaky corneal ulcer areas from fluorescein staining images. The ulcer area was segmented within the cornea by employing a joint method of Otsu and Gaussian Mixture Modeling (GMM). In the GMM based segmentation, the total number of Gaussians was determined intelligently using an information theory based algorithm. And the fluorescein staining images were processed in the HSV color model rather than the original RGB color model, aiming to improve the segmentation results' robustness and accuracy. In the Otsu based segmentation, the images were processed in the grayscale space with Gamma correction being conducted before the Otsu binarization. Afterwards, morphological operations and median filtering were employed to further improve the Otsu segmentation result. The GMM and Otsu segmentation results were then intersected, for which post-processing was conducted by identifying and filling holes through a fast algorithm using priority queues of pixels. The proposed pipeline has been validated on a total of 150 clinical images. Accurate ulcer segmentation results have been obtained, with the mean Dice Similarity Coefficient (DSC) being 0.88 when comparing the automatic segmentation result with the manually-delineated gold standard. For images in the RGB color space, the mean DSC was 0.83, being much lower than that of the images in the HSV color space.

In this paper, an adaptive edge detection method (Canny operator and Otsu threshold selection based adaptive edge detection method - COAED) is proposed. The COAED method combines a new hybrid filter with Canny operator to avoid the conflict of Canny operator between noise removing and edge locating, and uses Otsu threshold selection method to determine Dual-threshold of Canny operator adaptively. The new hybrid filter firstly judges whether the pixel is polluted by impulse noise, and then uses a corresponding filter to process the current pixel. A median filter is used if the pixel is thought to be impulse noise; otherwise an improved mean filter is selected to weaken the Gaussian noise. After the image is smoothed by the hybrid filter, a Canny operator with small Gaussian variance is used to extract edge. Because only part of Gaussian noise remains, Canny operator with small Gaussian variance can suppress the noise and preserve the edge effectively. Using the gauge image polluted by hybrid noise as experiment object, the performance of COAED method is evaluated qualitatively and quantitatively. Experimental results show that the COAED method is superior to Canny operator.

ObjectiveTo explain adult-type diffuse gliomas heterogeneity through diffusion kurtosis imaging-based habitat characteristics and develop and validate a comprehensive model for predicting isocitrate dehydrogenase (IDH) status.Materials and methodsIn this prospective secondary analysis, 103 participants (mean age, 52 years; range, 21-77; 54 [52%] male) pathologically diagnosed with adult-type diffuse gliomas were enrolled between June 2018 and February 2022. The Otsu method was used to generate habitat maps with mean diffusivity (MD) and mean kurtosis (MK) for a total of 4 subhabitats containing 16 habitat features. Habitat heatmaps were created based on the Pearson correlation coefficient. The Habitat imAging aNd clinicraD INtegrated prEdiction SyStem (HANDINESS) was created by combining clinical features, conventional MRI morphological features, and habitat image features. ROC, calibration curve, and decision curve analyses were used to select the optimal model after 32 pipelines for model training and validation.ResultsIn the restricted diffusion and high-density subhabitat, MK was highly correlated with MD (R2 = 0.999), volume (0.608) and percentage of volume (0.663), and this region had the highest MK value (P<.001). The unrestricted diffusion and low-density subhabitat had the highest MD value (P<.001). When MK was less than the Otsu threshold, there was still a difference between restricted diffusion and low-density and unrestricted diffusion and low-density subhabitats (P<.01). The HANDINESS enabled more accurate prediction of the IDH status in the training (AUC=0.951 [0.902-0.987]) and internal validation cohorts (0.938 [0.881-0.949]). AUC values for single-modality models and independent factors ranged from 0.593 to 0.916. Calibration and decision curve analyses showed that the HANDINESS demonstrated a high level of clinical applicability and predictive consistency.ConclusionDiffusion kurtosis imaging-based habitat analysis provides additional important information on microscopic tumor spatial heterogeneity. The HANDINESS has higher diagnostic performance and robustness than single-modality models.

In this paper, we comprehensively evaluated clinical application of local robust-region based algorithms to delineate the brain target volumes in radiation therapy treatment planning. Localized region based algorithms can optimize processing time of manual target tumor delineation and have perfect correlation with manual delineation defined by oncologist due to high deformability. Accordingly, they can receive much attention in radiation therapy treatment planning. Firstly, clinical target volumes (CTVs) of 135 slices in 18 patients were manually defined by two oncologists and the average of these contours considered as references in order to compare with semi-automatic results from different four algorithms. Then, four localized region based algorithms named Localizing Region Based Active Contour (LRBAC), Local Chan-Vese Model (LCV), Local Region Chan-Vese Model (LRCV) and Local Gaussian Distribution Fitting (LGDF) were applied to outline CTVs. Finally, comparisons between semiautomatic results and baselines were done according to three different metric criteria: Dice coefficient, Hausdorff distance, and mean absolute distance. Manual delineation processing times of target tumors were also performed. Our result showed that LCV has advantage over other algorithms in terms of the processing time and afterward LRCV is the second fastest method. LRBAC was the second slowest technique; however, we found that processing speed in LRBAC can be almost doubled by replacing the time-consuming re-initialization process with energy penalizing term. Accordingly, due to high accuracy performance of LRBAC algorithm, it can be concluded that the modified version of LRBAC has the best performance in brain target volumes in radiation therapy treatment planning among other localized algorithms in terms of speed and accuracy.KeywordsBrain Tumor TargetSegmentationregionbasedactive contourTreatment Planning

Accurate segmentation of ovarian cancer metastases is clinically useful to evaluate tumor growth and determine follow-up treatment. We present a region-based level set algorithm with localization constraints to segment ovarian cancer metastases. Our approach is established on a representative region-based level set, Chan-Vese model, in which an active contour is driven by region competition. To reduce over-segmentation, we constrain the level set propagation within a narrow image band by embedding a dynamic localization function. The metastasis intensity prior is also estimated from image regions within the level set initialization. The localization function and intensity prior force the level set to stop at the desired metastasis boundaries. Our approach was validated on 19 ovarian cancer metastases with radiologist-labeled ground-truth on contrast-enhanced CT scans from 15 patients. The comparison between our algorithm and geodesic active contour indicated that the volume overlap was 75&plusmn;10% vs. 56&plusmn;6%, the Dice coefficient was 83&plusmn;8% vs. 63&plusmn;8%, and the average surface distance was 2.2&plusmn;0.6mm vs. 4.4&plusmn;0.9mm. Experimental results demonstrated that our algorithm outperformed traditional level set algorithms.

Ra-223 is a promising radionuclide for the treatment of skeletal metastases in castration-resistant prostate cancer patients. This study aims to estimate the lower limits for feasible Ra-223 single-photon emission computerized tomography (SPECT) imaging using a Monte Carlo simulation study. The SPECT images were produced on a homemade code: the Monte Carlo simulation of electrons and photons for SPECT (MCEP-SPECT). The National Electrical Manufacturers Association (NEMA) phantom with six hot spheres of diameters of 37, 28, 22, 17, 13, and 10 mm installed inside, was used. The background activity concentration was 0.6 kBq/mL, and the ratios of hot concentrations to the background (R HB) were 25, 20, 15, 10, and 5. When R HB was 15, the background concentrations of 1.5, 0.9, 0.3, and 0.15 kBq/mL were also tested. The energy window was 84 keV±10%. The number of projections was 60/360°, and the acquisition time was 60 s per projection. Two kinds of collimators: middle-energy general-purpose (MEGP) and high-energy general-purpose (HEGP), were examined. The SPECT images were evaluated based on two quantitative indexes: contrast-to-noise ratio (CNR) for detectability and contrast recovery coefficient (CRC) for quantitative accuracy. The CRC for the HEGP collimator was 35-40%, while the CRC for the MEGP collimator was 25-30%. The CNRs for the MEGP collimator were larger than those for the HEGP collimator. The CNRs of the hot spheres with diameters less than 22 mm were lower than 5.0 for both collimators, when R HB and the background concentration were 15 and 0.6 kBq/mL, respectively. Based on the obtained results, it was estimated that the lower limit of R HB for the detection of the hot sphere with a diameter of 37 mm would be approximately 20 if the background concentration is 0.05 kBq/mL. The MEGP collimator is superior in terms of detectability, while the HEGP collimator is superior in terms of quantitative accuracy. When the lesion size is small, the MEGP collimator may be favorable. Based on these results, the estimated lower limit of the activity concentration would be approximately 1 kBq/mL if the background concentration is 0.05 kBq/mL for a large lesion.

Background: The quality of positron emission tomography/computed tomography (PET/CT) images plays an important role in tumor detection. This imaging method often yields poor-quality images of overweight patients due to the high level of noise, originating from scattering and photon attenuation. Objectives: The point spread function (PSF) is mostly used to enhance the spatial resolution and signal-to-noise ratio (SNR); however, it is known to increase the edge artifacts. The time-of-flight (TOF) principle can reduce edge artifacts in PSF modeling and improve lesion detection, especially in the thorax. The present study aimed to assess these two new techniques by applying different reconstruction parameters. Materials and Methods: An in-house phantom with an inner diameter of 35 cm was used for the simulation of overweight patients. Lesion-to-background ratios (LBRs) of 2: 1 and 8: 1, as well as background activity concentrations of 3 and 5 kBq/cc, were considered in this study. The list-mode data were reconstructed with various reconstruction protocols, numbers of subsets, and filter sizes. Quantitative analyses, including the coefficient of variation (COV), SNR, and recovery coefficient (RC), were also carried out. Moreover, box-and-whisker plots were performed. Results: At LBR of 2: 1, by changing the protocol from ordered subset expectation maximization (OSEM) to OSEM + PSF + TOF, the median value of SNR for 13-mm lesions (37 mm) increased by 39.25% and 53.45% (42.22% and 56.21%), at background activity concentrations of 3 and 5 kBq/cc respectively. However, at LBR of 8: 1, the corresponding values were 33.22% and 48.94% (40.22% and 52.15%) at background activity concentrations of 3 and 5 kBq/cc respectively. Conclusion: The TOF protocols were strongly recommended for both background activity concentrations at LBR of 2: 1 and for the low background activity concentration at LBR of 8: 1, especially when using smaller filter sizes. Moreover, subset numbers of 18 and 24 were appropriate for all protocols. However, a smaller subset number was suitable when a low background activity concentration and a smaller filter size were applied, especially at a lower LBR.

Water resources are an important component of ecosystem services. During long periods of cloudiness and precipitation, when a ground-based sample is not available, the water bodies are detected from satellite SAR (synthetic-aperture radar) data using threshold methods (e.g., Otsu and Kittler–Illingworth). However, such methods do not enable to obtain the correct threshold value for the backscattering coefficient (σ0) of relatively small water areas in the image. The paper proposes and substantiates a method for the mapping of the surface of water bodies, which makes it possible to correctly identify water bodies, even in “water”/“land” class imbalance situations. The method operates on a principle of maximum compliance of the resulting SAR water mask with a given reference water mask. Therefore, the method enables the exploration of the possibilities of searching and choosing the optimal parameters (polarization and speckle filtering), which provide the maximum quality of SAR water mask. The method was applied for mapping natural and industrial water bodies in the Pohjois-Pohjanmaa region (North Ostrobothnia), Finland, using Sentinel-1A and -1B ground range detected (GRD) data (ascending and descending orbits) in 2018–2021. Reference water masks were generated based on optical spectral indices derived from Sentinel-2A and -2B data. The polarization and speckle filtering parameters were chosen since they provide the most accurate σ0 threshold (on average for all observations above 0.9 according to the Intersection over Union criterion) and are resistant to random fluctuations. If a reference water mask is available, the proposed method is more accurate than the Otsu method. Without a reference mask, the σ0 threshold is calculated as an average of thresholds obtained from previous observations. In this case, the proposed method is as good in accuracy as the Otsu method. It is shown that the proposed method enables the identification of surface water bodies under significant class imbalance conditions, such as when the water surface covers only a fraction of a percent of the area under study.