Displaying all 7 publications

Abstract:
Sort:
  1. Pszczolkowski S, Manzano-Patrón JP, Law ZK, Krishnan K, Ali A, Bath PM, et al.
    Eur Radiol, 2021 Oct;31(10):7945-7959.
    PMID: 33860831 DOI: 10.1007/s00330-021-07826-9
    OBJECTIVES: To test radiomics-based features extracted from noncontrast CT of patients with spontaneous intracerebral haemorrhage for prediction of haematoma expansion and poor functional outcome and compare them with radiological signs and clinical factors.

    MATERIALS AND METHODS: Seven hundred fifty-four radiomics-based features were extracted from 1732 scans derived from the TICH-2 multicentre clinical trial. Features were harmonised and a correlation-based feature selection was applied. Different elastic-net parameterisations were tested to assess the predictive performance of the selected radiomics-based features using grid optimisation. For comparison, the same procedure was run using radiological signs and clinical factors separately. Models trained with radiomics-based features combined with radiological signs or clinical factors were tested. Predictive performance was evaluated using the area under the receiver operating characteristic curve (AUC) score.

    RESULTS: The optimal radiomics-based model showed an AUC of 0.693 for haematoma expansion and an AUC of 0.783 for poor functional outcome. Models with radiological signs alone yielded substantial reductions in sensitivity. Combining radiomics-based features and radiological signs did not provide any improvement over radiomics-based features alone. Models with clinical factors had similar performance compared to using radiomics-based features, albeit with low sensitivity for haematoma expansion. Performance of radiomics-based features was boosted by incorporating clinical factors, with time from onset to scan and age being the most important contributors for haematoma expansion and poor functional outcome prediction, respectively.

    CONCLUSION: Radiomics-based features perform better than radiological signs and similarly to clinical factors on the prediction of haematoma expansion and poor functional outcome. Moreover, combining radiomics-based features with clinical factors improves their performance.

    KEY POINTS: • Linear models based on CT radiomics-based features perform better than radiological signs on the prediction of haematoma expansion and poor functional outcome in the context of intracerebral haemorrhage. • Linear models based on CT radiomics-based features perform similarly to clinical factors known to be good predictors. However, combining these clinical factors with radiomics-based features increases their predictive performance.

  2. Kok YE, Pszczolkowski S, Law ZK, Ali A, Krishnan K, Bath PM, et al.
    Radiol Artif Intell, 2022 Nov;4(6):e220096.
    PMID: 36523645 DOI: 10.1148/ryai.220096
    This study evaluated deep learning algorithms for semantic segmentation and quantification of intracerebral hemorrhage (ICH), perihematomal edema (PHE), and intraventricular hemorrhage (IVH) on noncontrast CT scans of patients with spontaneous ICH. Models were assessed on 1732 annotated baseline noncontrast CT scans obtained from the Tranexamic Acid for Hyperacute Primary Intracerebral Haemorrhage (ie, TICH-2) international multicenter trial (ISRCTN93732214), and different loss functions using a three-dimensional no-new-U-Net (nnU-Net) were examined to address class imbalance (30% of participants with IVH in dataset). On the test cohort (n = 174, 10% of dataset), the top-performing models achieved median Dice similarity coefficients of 0.92 (IQR, 0.89-0.94), 0.66 (0.58-0.71), and 1.00 (0.87-1.00), respectively, for ICH, PHE, and IVH segmentation. U-Net-based networks showed comparable, satisfactory performances on ICH and PHE segmentations (P > .05), but all nnU-Net variants achieved higher accuracy than the Brain Lesion Analysis and Segmentation Tool for CT (BLAST-CT) and DeepLabv3+ for all labels (P < .05). The Focal model showed improved performance in IVH segmentation compared with the Tversky, two-dimensional nnU-Net, U-Net, BLAST-CT, and DeepLabv3+ models (P < .05). Focal achieved concordance values of 0.98, 0.88, and 0.99 for ICH, PHE, and ICH volumes, respectively. The mean volumetric differences between the ground truth and prediction were 0.32 mL (95% CI: -8.35, 9.00), 1.14 mL (-9.53, 11.8), and 0.06 mL (-1.71, 1.84), respectively. In conclusion, U-Net-based networks provide accurate segmentation on CT images of spontaneous ICH, and Focal loss can address class imbalance. International Clinical Trials Registry Platform (ICTRP) no. ISRCTN93732214 Supplemental material is available for this article. © RSNA, 2022 Keywords: Head/Neck, Brain/Brain Stem, Hemorrhage, Segmentation, Quantification, Convolutional Neural Network (CNN), Deep Learning Algorithms, Machine Learning Algorithms.
  3. Pszczolkowski S, Law ZK, Gallagher RG, Meng D, Swienton DJ, Morgan PS, et al.
    Comput Biol Med, 2019 Mar;106:126-139.
    PMID: 30711800 DOI: 10.1016/j.compbiomed.2019.01.022
    BACKGROUND: Spontaneous intracerebral haemorrhage (SICH) is a common condition with high morbidity and mortality. Segmentation of haematoma and perihaematoma oedema on medical images provides quantitative outcome measures for clinical trials and may provide important markers of prognosis in people with SICH.

    METHODS: We take advantage of improved contrast seen on magnetic resonance (MR) images of patients with acute and early subacute SICH and introduce an automated algorithm for haematoma and oedema segmentation from these images. To our knowledge, there is no previously proposed segmentation technique for SICH that utilises MR images directly. The method is based on shape and intensity analysis for haematoma segmentation and voxel-wise dynamic thresholding of hyper-intensities for oedema segmentation.

    RESULTS: Using Dice scores to measure segmentation overlaps between labellings yielded by the proposed algorithm and five different expert raters on 18 patients, we observe that our technique achieves overlap scores that are very similar to those obtained by pairwise expert rater comparison. A further comparison between the proposed method and a state-of-the-art Deep Learning segmentation on a separate set of 32 manually annotated subjects confirms the proposed method can achieve comparable results with very mild computational burden and in a completely training-free and unsupervised way.

    CONCLUSION: Our technique can be a computationally light and effective way to automatically delineate haematoma and oedema extent directly from MR images. Thus, with increasing use of MR images clinically after intracerebral haemorrhage this technique has the potential to inform clinical practice in the future.

  4. Dineen RA, Pszczolkowski S, Flaherty K, Law ZK, Morgan PS, Roberts I, et al.
    BMJ Open, 2018 02 03;8(2):e019930.
    PMID: 29431141 DOI: 10.1136/bmjopen-2017-019930
    OBJECTIVES: To test whether administration of the antifibrinolytic drug tranexamic acid (TXA) in patients with spontaneous intracerebral haemorrhage (SICH) leads to increased prevalence of diffusion-weighted MRI-defined hyperintense ischaemic lesions (primary hypothesis) or reduced perihaematomal oedema volume, perihaematomal diffusion restriction and residual MRI-defined SICH-related tissue damage (secondary hypotheses).

    DESIGN: MRI substudy nested within the double-blind randomised controlled Tranexamic Acid for Hyperacute Primary Intracerebral Haemorrhage (TICH)-2 trial (ISRCTN93732214).

    SETTING: International multicentre hospital-based study.

    PARTICIPANTS: Eligible adults consented and randomised in the TICH-2 trial who were also able to undergo MRI scanning. To address the primary hypothesis, a sample size of n=280 will allow detection of a 10% relative increase in prevalence of diffusion-weighted imaging (DWI) hyperintense lesions in the TXA group with 5% significance, 80% power and 5% imaging data rejection.

    INTERVENTIONS: TICH-2 MRI substudy participants will undergo MRI scanning using a standardised protocol at day ~5 and day ~90 after randomisation. Clinical assessments, randomisation to TXA or placebo and participant follow-up will be performed as per the TICH-2 trial protocol.

    CONCLUSION: The TICH-2 MRI substudy will test whether TXA increases the incidence of new DWI-defined ischaemic lesions or reduces perihaematomal oedema or final ICH lesion volume in the context of SICH.

    ETHICS AND DISSEMINATION: The TICH-2 trial obtained ethical approval from East Midlands - Nottingham 2 Research Ethics Committee (12/EM/0369) and an amendment to allow the TICH-2 MRI sub study was approved in April 2015 (amendment number SA02/15). All findings will be published in peer-reviewed journals. The primary outcome results will also be presented at a relevant scientific meeting.

    TRIAL REGISTRATION NUMBER: ISRCTN93732214; Pre-results.

  5. Law ZK, Ali A, Krishnan K, Bischoff A, Appleton JP, Scutt P, et al.
    Stroke, 2020 01;51(1):121-128.
    PMID: 31735141 DOI: 10.1161/STROKEAHA.119.026128
    Background and Purpose- Blend, black hole, island signs, and hypodensities are reported to predict hematoma expansion in acute intracerebral hemorrhage. We explored the value of these noncontrast computed tomography signs in predicting hematoma expansion and functional outcome in our cohort of intracerebral hemorrhage. Methods- The TICH-2 (Tranexamic acid for IntraCerebral Hemorrhage-2) was a prospective randomized controlled trial exploring the efficacy and safety of tranexamic acid in acute intracerebral hemorrhage. Baseline and 24-hour computed tomography scans of trial participants were analyzed. Hematoma expansion was defined as an increase in hematoma volume of >33% or >6 mL on 24-hour computed tomography. Poor functional outcome was defined as modified Rankin Scale of 4 to 6 at day 90. Multivariable logistic regression was performed to identify predictors of hematoma expansion and poor functional outcome. Results- Of 2325 patients recruited, 2077 (89.3%) had valid baseline and 24-hour scans. Five hundred seventy patients (27.4%) had hematoma expansion while 1259 patients (54.6%) had poor functional outcome. The prevalence of noncontrast computed tomography signs was blend sign, 366 (16.1%); black hole sign, 414 (18.2%); island sign, 200 (8.8%); and hypodensities, 701 (30.2%). Blend sign (adjusted odds ratio [aOR] 1.53 [95% CI, 1.16-2.03]; P=0.003), black hole (aOR, 2.03 [1.34-3.08]; P=0.001), and hypodensities (aOR, 2.06 [1.48-2.89]; P<0.001) were independent predictors of hematoma expansion on multivariable analysis with adjustment for covariates. Black hole sign (aOR, 1.52 [1.10-2.11]; P=0.012), hypodensities (aOR, 1.37 [1.05-1.78]; P=0.019), and island sign (aOR, 2.59 [1.21-5.55]; P=0.014) were significant predictors of poor functional outcome. Tranexamic acid reduced the risk of hematoma expansion (aOR, 0.77 [0.63-0.94]; P=0.010), but there was no significant interaction between the presence of noncontrast computed tomography signs and benefit of tranexamic acid on hematoma expansion and functional outcome (P interaction all >0.05). Conclusions- Blend sign, black hole sign, and hypodensities predict hematoma expansion while black hole sign, hypodensities, and island signs predict poor functional outcome. Noncontrast computed tomography signs did not predict a better response to tranexamic acid. Clinical Trial Registration- URL: https://www.isrctn.com. Unique identifier: ISRCTN93732214.
  6. Pszczolkowski S, Sprigg N, Woodhouse LJ, Gallagher R, Swienton D, Law ZK, et al.
    JAMA Neurol, 2022 May 01;79(5):468-477.
    PMID: 35311937 DOI: 10.1001/jamaneurol.2022.0217
    IMPORTANCE: Hyperintense foci on diffusion-weighted imaging (DWI) that are spatially remote from the acute hematoma occur in 20% of people with acute spontaneous intracerebral hemorrhage (ICH). Tranexamic acid, a hemostatic agent that is under investigation for treating acute ICH, might increase DWI hyperintense lesions (DWIHLs).

    OBJECTIVE: To establish whether tranexamic acid compared with placebo increased the prevalence or number of remote cerebral DWIHLs within 2 weeks of ICH onset.

    DESIGN, SETTING, AND PARTICIPANTS: This prospective nested magnetic resonance imaging (MRI) substudy of a randomized clinical trial (RCT) recruited participants from the multicenter, double-blind, placebo-controlled, phase 3 RCT (Tranexamic Acid for Hyperacute Primary Intracerebral Hemorrhage [TICH-2]) from July 1, 2015, to September 30, 2017, and conducted follow-up to 90 days after participants were randomized to either the tranexamic acid or placebo group. Participants had acute spontaneous ICH and included TICH-2 participants who provided consent to undergo additional MRI scans for the MRI substudy and those who had clinical MRI data that were compatible with the brain MRI protocol of the substudy. Data analyses were performed on an intention-to-treat basis on January 20, 2020.

    INTERVENTIONS: The tranexamic acid group received 1 g in 100-mL intravenous bolus loading dose, followed by 1 g in 250-mL infusion within 8 hours of ICH onset. The placebo group received 0.9% saline within 8 hours of ICH onset. Brain MRI scans, including DWI, were performed within 2 weeks.

    MAIN OUTCOMES AND MEASURES: Prevalence and number of remote DWIHLs were compared between the treatment groups using binary logistic regression adjusted for baseline covariates.

    RESULTS: A total of 219 participants (mean [SD] age, 65.1 [13.8] years; 126 men [57.5%]) who had brain MRI data were included. Of these participants, 96 (43.8%) were randomized to receive tranexamic acid and 123 (56.2%) were randomized to receive placebo. No baseline differences in demographic characteristics and clinical or imaging features were found between the groups. There was no increase for the tranexamic acid group compared with the placebo group in DWIHL prevalence (20 of 96 [20.8%] vs 28 of 123 [22.8%]; odds ratio [OR], 0.71; 95% CI, 0.33-1.53; P = .39) or mean (SD) number of DWIHLs (1.75 [1.45] vs 1.81 [1.71]; mean difference [MD], -0.08; 95% CI, -0.36 to 0.20; P = .59). In an exploratory analysis, participants who were randomized within 3 hours of ICH onset or those with chronic infarcts appeared less likely to have DWIHLs if they received tranexamic acid. Participants with probable cerebral amyloid angiopathy appeared more likely to have DWIHLs if they received tranexamic acid.

    CONCLUSIONS AND RELEVANCE: This substudy of an RCT found no evidence of increased prevalence or number of remote DWIHLs after tranexamic acid treatment in acute ICH. These findings provide reassurance for ongoing and future trials that tranexamic acid for acute ICH is unlikely to induce cerebral ischemic events.

    TRIAL REGISTRATION: isrctn.org Identifier: ISRCTN93732214.

Filters
Contact Us

Please provide feedback to Administrator ([email protected])

External Links