Change search
Link to record
Permanent link

Direct link
BETA
Alternative names
Publications (10 of 35) Show all publications
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: 2019-05-23Bibliographically 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, Vol. 133, p. S208-S209Article in journal (Refereed) Published
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: 2019-07-29Bibliographically 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)2-s2.0-85036471005 (Scopus ID)
Note

QC 20171215

Available from: 2017-12-15 Created: 2017-12-15 Last updated: 2017-12-15Bibliographically approved
Wang, C. & Smedby, Ö. (2018). Automatic whole heart segmentation using deep learning and shape context. In: 8th International Workshop on Statistical Atlases and Computational Models of the Heart, STACOM 2017, Held in Conjunction with MICCAI 2017: . Paper presented at 8th International Workshop on Statistical Atlases and Computational Models of the Heart, STACOM 2017, Held in Conjunction with MICCAI 2017, Quebec City, Canada, 10 September 2017 through 14 September 2017 (pp. 242-249). Springer, 10663
Open this publication in new window or tab >>Automatic whole heart segmentation using deep learning and shape context
2018 (English)In: 8th International Workshop on Statistical Atlases and Computational Models of the Heart, STACOM 2017, Held in Conjunction with MICCAI 2017, Springer, 2018, Vol. 10663, p. 242-249Conference paper, Published paper (Refereed)
Abstract [en]

To assist 3D cardiac image analysis, we propose an automatic whole heart segmentation using a deep learning framework combined with shape context information that is encoded in volumetric shape models. The proposed processing pipeline consists of three major steps: scout segmentation with orthogonal 2D U-nets, shape context estimation and refining segmentation with U-net and shape context. The proposed method was evaluated using the MMWHS challenge data. Two sets of networks were trained separately for contrast-enhanced CT and MRI. On the 20 training datasets, using 5-fold cross-validation, the average Dice coefficients for the left ventricle, the right ventricle, the left atrium, the right atrium and the myocardium of the left ventricle were 0.895, 0.795, 0.847, 0.821, 0.807 for MRI and 0.935, 0.825, 0.908, 0.881, 0.879 for CT, respectively. Further improvement may be possible given more training data or advanced data augmentation strategy.

Place, publisher, year, edition, pages
Springer, 2018
Series
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), ISSN 0302-9743 ; 10663
Keywords
Deep learning, Fully convolutional network, Heart segmentation, Shape context, Statistic shape model
National Category
Medical and Health Sciences
Identifiers
urn:nbn:se:kth:diva-225494 (URN)10.1007/978-3-319-75541-0_26 (DOI)2-s2.0-85044467877 (Scopus ID)9783319755403 (ISBN)
Conference
8th International Workshop on Statistical Atlases and Computational Models of the Heart, STACOM 2017, Held in Conjunction with MICCAI 2017, Quebec City, Canada, 10 September 2017 through 14 September 2017
Funder
Swedish Heart Lung Foundation, 2016-0609Swedish Research Council, 2014-6153
Note

QC 20180406

Available from: 2018-04-06 Created: 2018-04-06 Last updated: 2018-04-06Bibliographically approved
Mahbod, A., Ellinger, I., Ecker, R., Smedby, Ö. & Wang, C. (2018). Breast Cancer Histological Image Classification Using Fine-Tuned Deep Network Fusion. In: 15th International Conference on Image Analysis and Recognition, ICIAR 2018: . Paper presented at 27 June 2018 through 29 June 2018 (pp. 754-762). Springer
Open this publication in new window or tab >>Breast Cancer Histological Image Classification Using Fine-Tuned Deep Network Fusion
Show others...
2018 (English)In: 15th International Conference on Image Analysis and Recognition, ICIAR 2018, Springer, 2018, p. 754-762Conference paper, Published paper (Refereed)
Abstract [en]

Breast cancer is the most common cancer type in women worldwide. Histological evaluation of the breast biopsies is a challenging task even for experienced pathologists. In this paper, we propose a fully automatic method to classify breast cancer histological images to four classes, namely normal, benign, in situ carcinoma and invasive carcinoma. The proposed method takes normalized hematoxylin and eosin stained images as input and gives the final prediction by fusing the output of two residual neural networks (ResNet) of different depth. These ResNets were first pre-trained on ImageNet images, and then fine-tuned on breast histological images. We found that our approach outperformed a previous published method by a large margin when applied on the BioImaging 2015 challenge dataset yielding an accuracy of 97.22%. Moreover, the same approach provided an excellent classification performance with an accuracy of 88.50% when applied on the ICIAR 2018 grand challenge dataset using 5-fold cross validation.

Place, publisher, year, edition, pages
Springer, 2018
Keywords
Breast cancer, Classification, Deep learning, Histological images, Biopsy, Classification (of information), Diseases, Image analysis, Medical imaging, Automatic method, Breast biopsies, Classification performance, Cross validation, Grand Challenge, Histological evaluation, Image classification
National Category
Computer Vision and Robotics (Autonomous Systems)
Identifiers
urn:nbn:se:kth:diva-236392 (URN)10.1007/978-3-319-93000-8_85 (DOI)2-s2.0-85049429428 (Scopus ID)9783319929996 (ISBN)
Conference
27 June 2018 through 29 June 2018
Note

QC 20181101

Available from: 2018-11-01 Created: 2018-11-01 Last updated: 2018-11-01Bibliographically approved
Brusini, I., Carneiro, M., Wang, C., Rubin, C.-J., Ring, H., Afonso, S., . . . Andersson, L. (2018). Changes in brain architecture are consistent with altered fear processing in domestic rabbits. Proceedings of the National Academy of Sciences of the United States of America, 115(28), 7380-7385
Open this publication in new window or tab >>Changes in brain architecture are consistent with altered fear processing in domestic rabbits
Show others...
2018 (English)In: Proceedings of the National Academy of Sciences of the United States of America, ISSN 0027-8424, E-ISSN 1091-6490, Vol. 115, no 28, p. 7380-7385Article in journal (Refereed) Published
Abstract [en]

The most characteristic feature of domestic animals is their change in behavior associated with selection for tameness. Here we show, using high-resolution brain magnetic resonance imaging in wild and domestic rabbits, that domestication reduced amygdala volume and enlarged medial prefrontal cortex volume, supporting that areas driving fear have lost volume while areas modulating negative affect have gained volume during domestication. In contrast to the localized gray matter alterations, white matter anisotropy was reduced in the corona radiata, corpus callosum, and the subcortical white matter. This suggests a compromised white matter structural integrity in projection and association fibers affecting both afferent and efferent neural flow, consistent with reduced neural processing. We propose that compared with their wild ancestors, domestic rabbits are less fearful and have an attenuated flight response because of these changes in brain architecture.

Place, publisher, year, edition, pages
National Academy of Sciences, 2018
Keywords
rabbit, domestication, brain morphology, magnetic resonance imaging, fear
National Category
Neurosciences
Identifiers
urn:nbn:se:kth:diva-232773 (URN)10.1073/pnas.1801024115 (DOI)000438050900076 ()29941556 (PubMedID)2-s2.0-85049643502 (Scopus ID)
Funder
Knut and Alice Wallenberg FoundationSwedish Research CouncilThe Swedish Brain Foundation
Note

QC 20180807

Available from: 2018-08-07 Created: 2018-08-07 Last updated: 2019-06-18Bibliographically approved
Pavoni, M., Chang, Y., Park, S.-H. & Smedby, Ö. (2018). Convolutional neural network-based image enhancement for x-ray percutaneous coronary intervention. Journal of Medical Imaging, 5(2), Article ID 024006.
Open this publication in new window or tab >>Convolutional neural network-based image enhancement for x-ray percutaneous coronary intervention
2018 (English)In: Journal of Medical Imaging, ISSN 2329-4302, E-ISSN 2329-4310, Vol. 5, no 2, article id 024006Article in journal (Refereed) Published
Abstract [en]

Percutaneous coronary intervention (PCI) uses x-ray images, which may give high radiation dose and high concentrations of contrast media, leading to the risk of radiation-induced injury and nephropathy. These drawbacks can be reduced by using lower doses of x-rays and contrast media, with the disadvantage of noisier PCI images with less contrast. Vessel-edge-preserving convolutional neural networks (CNN) were designed to denoise simulated low x-ray dose PCI images, created by adding artificial noise to high-dose images. Objective functions of the designed CNNs have been optimized to achieve an edge-preserving effect of vessel walls, and the results of the proposed objective functions were evaluated qualitatively and quantitatively. Finally, the proposed CNN-based method was compared with two state-of-the-art denoising methods: K-SVD and block-matching and 3D filtering. The results showed promising performance of the proposed CNN-based method for PCI image enhancement with interesting capabilities of CNNs for real-time denoising and contrast enhancement tasks.

Place, publisher, year, edition, pages
Elsevier, 2018
National Category
Computer Vision and Robotics (Autonomous Systems) Radiology, Nuclear Medicine and Medical Imaging
Identifiers
urn:nbn:se:kth:diva-232924 (URN)10.1117/1.JMI.5.2.024006 (DOI)000439291500031 ()29963578 (PubMedID)2-s2.0-85049396885 (Scopus ID)
Note

QC 20180808

Available from: 2018-08-08 Created: 2018-08-08 Last updated: 2018-11-23Bibliographically approved
Klintström, E., Klintström, B., Pahr, D., Brismar, T. B., Smedby, Ö. & Moreno, R. (2018). Direct estimation of human trabecular bone stiffness using cone beam computed tomography. Oral surgery, oral medicine, oral pathology and oral radiology, 126(1), 72-82
Open this publication in new window or tab >>Direct estimation of human trabecular bone stiffness using cone beam computed tomography
Show others...
2018 (English)In: Oral surgery, oral medicine, oral pathology and oral radiology, ISSN 2212-4403, E-ISSN 2212-4411, Vol. 126, no 1, p. 72-82Article in journal (Refereed) Published
Abstract [en]

Objectives. The aim of this study was to evaluate the possibility of estimating the biomechanical properties of trabecular bone through finite element simulations by using dental cone beam computed tomography data. Study Design. Fourteen human radius specimens were scanned in 3 cone beam computed tomography devices: 3-D Accuitomo 80 (J. Morita MFG., Kyoto, Japan), NewTom 5 G (QR Verona, Verona, Italy), and Verity (Planmed, Helsinki, Finland). The imaging data were segmented by using 2 different methods. Stiffness (Young modulus), shear moduli, and the size and shape of the stiffness tensor were studied. Corresponding evaluations by using micro-CT were regarded as the reference standard. Results. The 3-D Accuitomo 80 (J. Morita MFG., Kyoto, Japan) showed good performance in estimating stiffness and shear moduli but was sensitive to the choice of segmentation method. Newtom 5 G (QR Verona, Verona, Italy) and Verity (Planmed, Helsinki, Finland) yielded good correlations, but they were not as strong as Accuitomo 80 U. Morita MFG., Kyoto, Japan). The cone beam computed tomography devices overestimated both stiffness and shear compared with the micro-CT estimations. Conclusions. Finite element-based calculations of biomechanics from cone beam computed tomography data are feasible, with strong correlations for the Accuitomo 80 scanner a. Morita MFG., Kyoto, Japan) combined with an appropriate segmentation method. Such measurements might be useful for predicting implant survival by in vivo estimations of bone properties.

Place, publisher, year, edition, pages
ELSEVIER SCIENCE INC, 2018
National Category
Medical Image Processing
Identifiers
urn:nbn:se:kth:diva-232242 (URN)10.1016/j.oooo.2018.03.014 (DOI)000436514800015 ()29735401 (PubMedID)2-s2.0-85046658791 (Scopus ID)
Note

QC 20180720

Available from: 2018-07-20 Created: 2018-07-20 Last updated: 2018-07-20Bibliographically approved
Buizza, G., Toma-Dasu, I., Lazzeroni, M., Paganelli, C., Riboldi, M., Chang, Y. J., . . . Wang, C. (2018). Early tumor response prediction for lung cancer patients using novel longitudinal pattern features from sequential PET/CT image scans. Physica medica (Testo stampato), 54, 21-29
Open this publication in new window or tab >>Early tumor response prediction for lung cancer patients using novel longitudinal pattern features from sequential PET/CT image scans
Show others...
2018 (English)In: Physica medica (Testo stampato), ISSN 1120-1797, E-ISSN 1724-191X, Vol. 54, p. 21-29Article in journal (Refereed) Published
Abstract [en]

Purpose: A new set of quantitative features that capture intensity changes in PET/CT images over time and space is proposed for assessing the tumor response early during chemoradiotherapy. The hypothesis whether the new features, combined with machine learning, improve outcome prediction is tested. Methods: The proposed method is based on dividing the tumor volume into successive zones depending on the distance to the tumor border. Mean intensity changes are computed within each zone, for CT and PET scans separately, and used as image features for tumor response assessment. Doing so, tumors are described by accounting for temporal and spatial changes at the same time. Using linear support vector machines, the new features were tested on 30 non-small cell lung cancer patients who underwent sequential or concurrent chemoradiotherapy. Prediction of 2-years overall survival was based on two PET-CT scans, acquired before the start and during the first 3 weeks of treatment. The predictive power of the newly proposed longitudinal pattern features was compared to that of previously proposed radiomics features and radiobiological parameters. Results: The highest areas under the receiver operating characteristic curves were 0.98 and 0.93 for patients treated with sequential and concurrent chemoradiotherapy, respectively. Results showed an overall comparable performance with respect to radiomics features and radiobiological parameters. Conclusions: A novel set of quantitative image features, based on underlying tumor physiology, was computed from PET/CT scans and successfully employed to distinguish between early responders and non-responders to chemoradiotherapy.

Place, publisher, year, edition, pages
ELSEVIER SCI LTD, 2018
Keywords
Early tumor response, Feature extraction, Non-small cell lung cancer, PET/CT
National Category
Radiology, Nuclear Medicine and Medical Imaging
Identifiers
urn:nbn:se:kth:diva-238132 (URN)10.1016/j.ejmp.2018.09.003 (DOI)000447271300003 ()30337006 (PubMedID)2-s2.0-85053799575 (Scopus ID)
Funder
Swedish Childhood Cancer Foundation, MT2016-0016The Swedish Brain FoundationEU, FP7, Seventh Framework Programme, ARTFORCEEU, European Research Council, ERC-ADG-2015EU, Horizon 2020, 733008
Note

QC 20181113

Available from: 2018-11-13 Created: 2018-11-13 Last updated: 2018-11-13Bibliographically approved
Organisations
Identifiers
ORCID iD: ORCID iD iconorcid.org/0000-0002-7750-1917

Search in DiVA

Show all publications