Change search
Link to record
Permanent link

Direct link
BETA
Smedby, Örjan, ProfessorORCID iD iconorcid.org/0000-0002-7750-1917
Alternative names
Publications (10 of 59) Show all publications
Mahbod, A., Schaefer, G., Ellinger, I., Ecker, R., Smedby, Ö. & Wang, C. (2019). A Two-Stage U-Net Algorithm for Segmentation of Nuclei in H&E-Stained Tissues. In: Constantino Carlos Reyes-Aldasoro, Andrew Janowczyk, Mitko Veta, Peter Bankhead, Korsuk Sirinukunwattana (Ed.), Digital Pathology: 15th European Congress, ECDP 2019, Warwick, UK, April 10–13, 2019, Proceedings. Paper presented at 15th European Congress on Digital Pathology, ECDP 2019, Warwick, United Kingdom 10-13 April 2019 (pp. 75-82). Springer Verlag
Open this publication in new window or tab >>A Two-Stage U-Net Algorithm for Segmentation of Nuclei in H&E-Stained Tissues
Show others...
2019 (English)In: Digital Pathology: 15th European Congress, ECDP 2019, Warwick, UK, April 10–13, 2019, Proceedings / [ed] Constantino Carlos Reyes-Aldasoro, Andrew Janowczyk, Mitko Veta, Peter Bankhead, Korsuk Sirinukunwattana, Springer Verlag , 2019, p. 75-82Conference paper, Published paper (Refereed)
Abstract [en]

Nuclei segmentation is an important but challenging task in the analysis of hematoxylin and eosin (H&E)-stained tissue sections. While various segmentation methods have been proposed, machine learning-based algorithms and in particular deep learning-based models have been shown to deliver better segmentation performance. In this work, we propose a novel approach to segment touching nuclei in H&E-stained microscopic images using U-Net-based models in two sequential stages. In the first stage, we perform semantic segmentation using a classification U-Net that separates nuclei from the background. In the second stage, the distance map of each nucleus is created using a regression U-Net. The final instance segmentation masks are then created using a watershed algorithm based on the distance maps. Evaluated on a publicly available dataset containing images from various human organs, the proposed algorithm achieves an average aggregate Jaccard index of 56.87%, outperforming several state-of-the-art algorithms applied on the same dataset.

Place, publisher, year, edition, pages
Springer Verlag, 2019
Series
Lecture Notes in Computer Science, ISSN 0302-9743, E-ISSN 1611-3349
Keywords
Deep learning, Digital pathology, Nuclei segmentation, Tissue analysis, U-Net, Machine learning, Pathology, Semantics, Tissue, Digital pathologies, Learning Based Models, Segmentation methods, Segmentation performance, Semantic segmentation, State-of-the-art algorithms, Image segmentation
National Category
Computer Sciences
Identifiers
urn:nbn:se:kth:diva-262448 (URN)10.1007/978-3-030-23937-4_9 (DOI)2-s2.0-85069146581 (Scopus ID)9783030239367 (ISBN)
Conference
15th European Congress on Digital Pathology, ECDP 2019, Warwick, United Kingdom 10-13 April 2019
Note

QC 20191021

Available from: 2019-10-21 Created: 2019-10-21 Last updated: 2019-10-21Bibliographically approved
Bendazzoli, S., Brusini, I., Damberg, P., Smedby, Ö., Andersson, L. & Wang, C. (2019). Automatic rat brain segmentation from MRI using statistical shape models and random forest. In: Angelini, ED Landman, BA (Ed.), MEDICAL IMAGING 2019: IMAGE PROCESSING. Paper presented at Conference on Medical Imaging: Image Processing, FEB 19-21, 2019, San Diego, CA. SPIE-INT SOC OPTICAL ENGINEERING, Article ID 1094920.
Open this publication in new window or tab >>Automatic rat brain segmentation from MRI using statistical shape models and random forest
Show others...
2019 (English)In: MEDICAL IMAGING 2019: IMAGE PROCESSING / [ed] Angelini, ED Landman, BA, SPIE-INT SOC OPTICAL ENGINEERING , 2019, article id 1094920Conference paper, Published paper (Refereed)
Abstract [en]

In MRI neuroimaging, the shimming procedure is used before image acquisition to correct for inhomogeneity of the static magnetic field within the brain. To correctly adjust the field, the brain's location and edges must first be identified from quickly-acquired low resolution data. This process is currently carried out manually by an operator, which can be time-consuming and not always accurate. In this work, we implement a quick and automatic technique for brain segmentation to be potentially used during the shimming. Our method is based on two main steps. First, a random forest classifier is used to get a preliminary segmentation from an input MRI image. Subsequently, a statistical shape model of the brain, which was previously generated from ground-truth segmentations, is fitted to the output of the classifier to obtain a model-based segmentation mask. In this way, a-priori knowledge on the brain's shape is included in the segmentation pipeline. The proposed methodology was tested on low resolution images of rat brains and further validated on rabbit brain images of higher resolution. Our results suggest that the present method is promising for the desired purpose in terms of time efficiency, segmentation accuracy and repeatability. Moreover, the use of shape modeling was shown to be particularly useful when handling low-resolution data, which could lead to erroneous classifications when using only machine learning-based methods.

Place, publisher, year, edition, pages
SPIE-INT SOC OPTICAL ENGINEERING, 2019
Series
Proceedings of SPIE, ISSN 0277-786X ; 10949
Keywords
brain MRI, image segmentation, shimming, random forest, statistical shape model
National Category
Radiology, Nuclear Medicine and Medical Imaging
Identifiers
urn:nbn:se:kth:diva-260221 (URN)10.1117/12.2512409 (DOI)000483012700090 ()2-s2.0-85068344757 (Scopus ID)978-1-5106-2546-4 (ISBN)
Conference
Conference on Medical Imaging: Image Processing, FEB 19-21, 2019, San Diego, CA
Note

QC 20190930

Available from: 2019-09-30 Created: 2019-09-30 Last updated: 2019-09-30Bibliographically approved
Astaraki, M., Wang, C., Buizza, G., Toma-Dasu, I., Lazzeroni, M. & Smedby, Ö. (2019). Early survival prediction in non-small cell lung cancer from PET/CT images using an intra-tumor partitioning method. Physica medica (Testo stampato), 60, 58-65
Open this publication in new window or tab >>Early survival prediction in non-small cell lung cancer from PET/CT images using an intra-tumor partitioning method
Show others...
2019 (English)In: Physica medica (Testo stampato), ISSN 1120-1797, E-ISSN 1724-191X, Vol. 60, p. 58-65Article in journal (Refereed) Published
Abstract [en]

Purpose: To explore prognostic and predictive values of a novel quantitative feature set describing intra-tumor heterogeneity in patients with lung cancer treated with concurrent and sequential chemoradiotherapy. Methods: Longitudinal PET-CT images of 30 patients with non-small cell lung cancer were analysed. To describe tumor cell heterogeneity, the tumors were partitioned into one to ten concentric regions depending on their sizes, and, for each region, the change in average intensity between the two scans was calculated for PET and CT images separately to form the proposed feature set. To validate the prognostic value of the proposed method, radiomics analysis was performed and a combination of the proposed novel feature set and the classic radiomic features was evaluated. A feature selection algorithm was utilized to identify the optimal features, and a linear support vector machine was trained for the task of overall survival prediction in terms of area under the receiver operating characteristic curve (AUROC). Results: The proposed novel feature set was found to be prognostic and even outperformed the radiomics approach with a significant difference (AUROC(sALop) = 0.90 vs. AUROC(radiomic) = 0.71) when feature selection was not employed, whereas with feature selection, a combination of the novel feature set and radiomics led to the highest prognostic values. Conclusion: A novel feature set designed for capturing intra-tumor heterogeneity was introduced. Judging by their prognostic power, the proposed features have a promising potential for early survival prediction.

Place, publisher, year, edition, pages
ELSEVIER SCI LTD, 2019
Keywords
Survival prediction, Treatment response, Radiomics, Tumor heterogeneity, LONG ER, 1988, BIOMETRICS, V44, P837
National Category
Radiology, Nuclear Medicine and Medical Imaging
Identifiers
urn:nbn:se:kth:diva-251338 (URN)10.1016/j.ejmp.2019.03.024 (DOI)000464560200009 ()31000087 (PubMedID)2-s2.0-85063364742 (Scopus ID)
Note

QC 20190523

Available from: 2019-05-23 Created: 2019-05-23 Last updated: 2019-10-09Bibliographically approved
Astaraki, M., Wang, C., Buizza, G., Toma-Dasu, I., Lazzeroni, M. & Smedby, Ö. (2019). Early survival prediction in non-small cell lung cancer with PET/CT size aware longitudinal pattern. Paper presented at 38th Annual Meeting of the European-Society-for-Radiotherapy-and-Oncology (ESTRO), APR 26-30, 2019, Milan, ITALY. Radiotherapy and Oncology, 133, S208-S209
Open this publication in new window or tab >>Early survival prediction in non-small cell lung cancer with PET/CT size aware longitudinal pattern
Show others...
2019 (English)In: Radiotherapy and Oncology, ISSN 0167-8140, E-ISSN 1879-0887, ISSN 0167-8140, Vol. 133, p. S208-S209Article in journal (Refereed) Published
Keywords
Oncology; Radiology, Nuclear Medicine & Medical Imaging
National Category
Medical and Health Sciences
Identifiers
urn:nbn:se:kth:diva-252991 (URN)10.1016/S0167-8140(19)30826-6 (DOI)000468315601037 ()
Conference
38th Annual Meeting of the European-Society-for-Radiotherapy-and-Oncology (ESTRO), APR 26-30, 2019, Milan, ITALY
Note

QC 20190729

Available from: 2019-07-29 Created: 2019-07-29 Last updated: 2019-09-30Bibliographically approved
Zhuang, X., Li, L., Payer, C., Štern, D., Urschler, M., Heinrich, M. P., . . . Yang, G. (2019). Evaluation of algorithms for Multi-Modality Whole Heart Segmentation: An open-access grand challenge.. Medical Image Analysis, 58, Article ID 101537.
Open this publication in new window or tab >>Evaluation of algorithms for Multi-Modality Whole Heart Segmentation: An open-access grand challenge.
Show others...
2019 (English)In: Medical Image Analysis, ISSN 1361-8415, E-ISSN 1361-8423, Vol. 58, article id 101537Article in journal (Refereed) Published
Abstract [en]

Knowledge of whole heart anatomy is a prerequisite for many clinical applications. Whole heart segmentation (WHS), which delineates substructures of the heart, can be very valuable for modeling and analysis of the anatomy and functions of the heart. However, automating this segmentation can be challenging due to the large variation of the heart shape, and different image qualities of the clinical data. To achieve this goal, an initial set of training data is generally needed for constructing priors or for training. Furthermore, it is difficult to perform comparisons between different methods, largely due to differences in the datasets and evaluation metrics used. This manuscript presents the methodologies and evaluation results for the WHS algorithms selected from the submissions to the Multi-Modality Whole Heart Segmentation (MM-WHS) challenge, in conjunction with MICCAI 2017. The challenge provided 120 three-dimensional cardiac images covering the whole heart, including 60 CT and 60 MRI volumes, all acquired in clinical environments with manual delineation. Ten algorithms for CT data and eleven algorithms for MRI data, submitted from twelve groups, have been evaluated. The results showed that the performance of CT WHS was generally better than that of MRI WHS. The segmentation of the substructures for different categories of patients could present different levels of challenge due to the difference in imaging and variations of heart shapes. The deep learning (DL)-based methods demonstrated great potential, though several of them reported poor results in the blinded evaluation. Their performance could vary greatly across different network structures and training strategies. The conventional algorithms, mainly based on multi-atlas segmentation, demonstrated good performance, though the accuracy and computational efficiency could be limited. The challenge, including provision of the annotated training data and the blinded evaluation for submitted algorithms on the test data, continues as an ongoing benchmarking resource via its homepage (www.sdspeople.fudan.edu.cn/zhuangxiahai/0/mmwhs/).

Keywords
Benchmark, Challenge, Multi-modality, Whole Heart Segmentation
National Category
Medical Image Processing
Identifiers
urn:nbn:se:kth:diva-258723 (URN)10.1016/j.media.2019.101537 (DOI)31446280 (PubMedID)2-s2.0-85070924419 (Scopus ID)
Note

QC 20190911

Available from: 2019-09-10 Created: 2019-09-10 Last updated: 2019-11-11Bibliographically approved
Kataria, B., Althén, J. N., Smedby, Ö., Persson, A., Sökjer, H. & Sandborg, M. (2019). Image quality and pathology assessment in CT Urography: when is the low-dose series sufficient?. BMC Medical Imaging, 19(1), Article ID 64.
Open this publication in new window or tab >>Image quality and pathology assessment in CT Urography: when is the low-dose series sufficient?
Show others...
2019 (English)In: BMC Medical Imaging, ISSN 1471-2342, E-ISSN 1471-2342, Vol. 19, no 1, article id 64Article in journal (Refereed) Published
Abstract [en]

Background Our aim was to compare CT images from native, nephrographic and excretory phases using image quality criteria as well as the detection of positive pathological findings in CT Urography, to explore if the radiation burden to the younger group of patients or patients with negative outcomes can be reduced. Methods This is a retrospective study of 40 patients who underwent a CT Urography examination on a 192-slice dual source scanner. Image quality was assessed for four specific renal image criteria from the European guidelines, together with pathological assessment in three categories: renal, other abdominal, and incidental findings without clinical significance. Each phase was assessed individually by three radiologists with varying experience using a graded scale. Certainty scores were derived based on the graded assessments. Statistical analysis was performed using visual grading regression (VGR). The limit for significance was set at p = 0.05. Results For visual reproduction of the renal parenchyma and renal arteries, the image quality was judged better for the nephrogram phase (p < 0.001), whereas renal pelvis/calyces and proximal ureters were better reproduced in the excretory phase compared to the native phase (p < 0.001). Similarly, significantly higher certainty scores were obtained in the nephrogram phase for renal parenchyma and renal arteries, but in the excretory phase for renal pelvis/calyxes and proximal ureters. Assessment of pathology in the three categories showed no statistically significant differences between the three phases. Certainty scores for assessment of pathology, however, showed a significantly higher certainty for renal pathology when comparing the native phase to nephrogram and excretory phase and a significantly higher score for nephrographic phase but only for incidental findings. Conclusion Visualisation of renal anatomy was as expected with each post-contrast phase showing favourable scores compared to the native phase. No statistically significant differences in the assessment of pathology were found between the three phases. The low-dose CT (LDCT) seems to be sufficient in differentiating between normal and pathological examinations. To reduce the radiation burden in certain patient groups, the LDCT could be considered a suitable alternative as a first line imaging method. However, radiologists should be aware of its limitations.

Place, publisher, year, edition, pages
BMC, 2019
Keywords
Computed tomography, Urography, Low-dose, Optimization, Image quality, Dose
National Category
Medical Image Processing
Identifiers
urn:nbn:se:kth:diva-257446 (URN)10.1186/s12880-019-0363-z (DOI)000480486200001 ()31399078 (PubMedID)2-s2.0-85070460822 (Scopus ID)
Note

QC 20190902

Available from: 2019-09-02 Created: 2019-09-02 Last updated: 2019-09-02Bibliographically approved
Wan, F., Smedby, Ö. & Wang, C. (2019). Simultaneous MR knee image segmentation and bias field correction using deep learning and partial convolution. In: Medical Imaging 2019: Image Processing. Paper presented at Medical Imaging 2019: Image Processing; San Diego; United States; 19 February 2019 through 21 February 2019. SPIE - International Society for Optical Engineering, 10949, Article ID 1094909.
Open this publication in new window or tab >>Simultaneous MR knee image segmentation and bias field correction using deep learning and partial convolution
2019 (English)In: Medical Imaging 2019: Image Processing, SPIE - International Society for Optical Engineering, 2019, Vol. 10949, article id 1094909Conference paper, Published paper (Refereed)
Abstract [en]

Intensity inhomogeneity is a great challenge for automated organ segmentation in magnetic resonance (MR) images. Many segmentation methods fail to deliver satisfactory results when the images are corrupted by a bias field. Although inhomogeneity correction methods exist, they often fail to remove the bias field completely in knee MR images. We present a new iterative approach that simultaneously predicts the segmentation mask of knee structures using a 3D U-net and estimates the bias field in 3D MR knee images using partial convolution operations. First, the test images run through a trained 3D U-net to generate a preliminary segmentation result, which is then fed to the partial convolution filter to create a preliminary estimation of the bias field using the segmented bone mask. Finally, the estimated bias field is then used to produce bias field corrected images as the new inputs to the 3D U-net. Through this loop, the segmentation results and bias field correction are iteratively improved. The proposed method was evaluated on 20 proton-density (PD)-weighted knee MRI scans with manually created segmentation ground truth using 10 fold cross-validation. In our preliminary experiments, the proposed methods outperformed conventional inhomogeneity-correction-plus-segmentation setup in terms of both segmentation accuracy and speed.

Place, publisher, year, edition, pages
SPIE - International Society for Optical Engineering, 2019
Series
Progress in Biomedical Optics and Imaging - Proceedings of SPIE, ISSN 1605-7422 ; 10949
National Category
Medical Image Processing
Identifiers
urn:nbn:se:kth:diva-258892 (URN)10.1117/12.2512950 (DOI)000483012700008 ()2-s2.0-85068319000 (Scopus ID)9781510625457 (ISBN)
Conference
Medical Imaging 2019: Image Processing; San Diego; United States; 19 February 2019 through 21 February 2019
Note

QC 20190913

Available from: 2019-09-11 Created: 2019-09-11 Last updated: 2019-09-26Bibliographically approved
Brusini, I., Jörgens, D., Smedby, Ö. & Moreno, R. (2019). Voxel-Wise Clustering of Tractography Data for Building Atlases of Local Fiber Geometry. In: : . Paper presented at International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2018; Granada; Spain; 20 September 2018 through 20 September 2018 (pp. 345-357).
Open this publication in new window or tab >>Voxel-Wise Clustering of Tractography Data for Building Atlases of Local Fiber Geometry
2019 (English)Conference paper, Published paper (Refereed)
Abstract [en]

This paper aims at proposing a method to generate atlases of white matter fibers’ geometry that consider local orientation and curvature of fibers extracted from tractography data. Tractography was performed on diffusion magnetic resonance images from a set of healthy subjects and each tract was characterized voxel-wise by its curvature and Frenet–Serret frame, based on which similar tracts could be clustered separately for each voxel and each subject. Finally, the centroids of the clusters identified in all subjects were clustered to create the final atlas. The proposed clustering technique showed promising results in identifying voxel-wise distributions of curvature and orientation. Two tractography algorithms (one deterministic and one probabilistic) were tested for the present work, obtaining two different atlases. A high agreement between the two atlases was found in several brain regions. This suggests that more advanced tractography methods might only be required for some specific regions in the brain. In addition, the probabilistic approach resulted in the identification of a higher number of fiber orientations in various white matter areas, suggesting it to be more adequate for investigating complex fiber configurations in the proposed framework as compared to deterministic tractography.

National Category
Medical Image Processing
Identifiers
urn:nbn:se:kth:diva-259768 (URN)10.1007/978-3-030-05831-9_27 (DOI)2-s2.0-85066883835 (Scopus ID)
Conference
International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2018; Granada; Spain; 20 September 2018 through 20 September 2018
Note

QC 20190923

Available from: 2019-09-23 Created: 2019-09-23 Last updated: 2019-09-23Bibliographically approved
Kataria, B., Althen, J. N., Smedby, Ö., Persson, A., Sokjer, H. & Sandborg, M. (2018). Assessment of image quality in abdominal CT: potential dose reduction with model-based iterative reconstruction. European Radiology, 28(6), 2464-2473
Open this publication in new window or tab >>Assessment of image quality in abdominal CT: potential dose reduction with model-based iterative reconstruction
Show others...
2018 (English)In: European Radiology, ISSN 0938-7994, E-ISSN 1432-1084, Vol. 28, no 6, p. 2464-2473Article in journal (Refereed) Published
Abstract [en]

To estimate potential dose reduction in abdominal CT by visually comparing images reconstructed with filtered back projection (FBP) and strengths of 3 and 5 of a specific MBIR. A dual-source scanner was used to obtain three data sets each for 50 recruited patients with 30, 70 and 100% tube loads (mean CTDIvol 1.9, 3.4 and 6.2 mGy). Six image criteria were assessed independently by five radiologists. Potential dose reduction was estimated with Visual Grading Regression (VGR). Comparing 30 and 70% tube load, improved image quality was observed as a significant strong effect of log tube load and reconstruction method with potential dose reduction relative to FBP of 22-47% for MBIR strength 3 (p < 0.001). For MBIR strength 5 no dose reduction was possible for image criteria 1 (liver parenchyma), but dose reduction between 34 and 74% was achieved for other criteria. Interobserver reliability showed agreement of 71-76% (kappa (w) 0.201-0.286) and intra-observer reliability of 82-96% (kappa (w) 0.525-0.783). MBIR showed improved image quality compared to FBP with positive correlation between MBIR strength and increasing potential dose reduction for all but one image criterion. aEuro cent MBIR's main advantage is its de-noising properties, which facilitates dose reduction. aEuro cent MBIR allows for potential dose reduction in relation to FBP. aEuro cent Visual Grading Regression (VGR) produces direct numerical estimates of potential dose reduction. aEuro cent MBIR strengths 3 and 5 dose reductions were 22-34 and 34-74%. aEuro cent MBIR strength 5 demonstrates inferior performance for liver parenchyma.

Place, publisher, year, edition, pages
SPRINGER, 2018
Keywords
Dose, Computed tomography, Iterative reconstruction, Abdomen, FBP
National Category
Radiology, Nuclear Medicine and Medical Imaging
Identifiers
urn:nbn:se:kth:diva-228459 (URN)10.1007/s00330-017-5113-4 (DOI)000431653200023 ()29368163 (PubMedID)
Note

QC 20180524

Available from: 2018-05-24 Created: 2018-05-24 Last updated: 2018-05-24Bibliographically approved
Mahbod, A., Chowdhury, M., Smedby, Ö. & Wang, C. (2018). Automatic brain segmentation using artificial neural networks with shape context. Pattern Recognition Letters, 101, 74-79
Open this publication in new window or tab >>Automatic brain segmentation using artificial neural networks with shape context
2018 (English)In: Pattern Recognition Letters, ISSN 0167-8655, E-ISSN 1872-7344, Vol. 101, p. 74-79Article in journal (Refereed) Published
Abstract [en]

Segmenting brain tissue from MR scans is thought to be highly beneficial for brain abnormality diagnosis, prognosis monitoring, and treatment evaluation. Many automatic or semi-automatic methods have been proposed in the literature in order to reduce the requirement of user intervention, but the level of accuracy in most cases is still inferior to that of manual segmentation. We propose a new brain segmentation method that integrates volumetric shape models into a supervised artificial neural network (ANN) framework. This is done by running a preliminary level-set based statistical shape fitting process guided by the image intensity and then passing the signed distance maps of several key structures to the ANN as feature channels, in addition to the conventional spatial-based and intensity-based image features. The so-called shape context information is expected to help the ANN to learn local adaptive classification rules instead of applying universal rules directly on the local appearance features. The proposed method was tested on a public datasets available within the open MICCAI grand challenge (MRBrainS13). The obtained average Dice coefficient were 84.78%, 88.47%, 82.76%, 95.37% and 97.73% for gray matter (GM), white matter (WM), cerebrospinal fluid (CSF), brain (WM + GM) and intracranial volume respectively. Compared with other methods tested on the same dataset, the proposed method achieved competitive results with comparatively shorter training time.

Place, publisher, year, edition, pages
Elsevier, 2018
National Category
Medical Image Processing
Identifiers
urn:nbn:se:kth:diva-219889 (URN)10.1016/j.patrec.2017.11.016 (DOI)000418101400011 ()2-s2.0-85036471005 (Scopus ID)
Note

QC 20171215

Available from: 2017-12-15 Created: 2017-12-15 Last updated: 2019-10-28Bibliographically approved
Organisations
Identifiers
ORCID iD: ORCID iD iconorcid.org/0000-0002-7750-1917

Search in DiVA

Show all publications

Profile pages