Rapid artefact removal and H&E-stained tissue segmentation

We present an innovative method for rapidly segmenting haematoxylin and eosin (H&E)-stained tissue in whole-slide images (WSIs) that eliminates a wide range of undesirable artefacts such as pen marks and scanning artefacts. Our method involves taking a single-channel representation of a low-magnification RGB overview of the WSI in which the pixel values are bimodally distributed such that H&E-stained tissue is easily distinguished from both background and a wide variety of artefacts. We demonstrate our method on 30 WSIs prepared from a wide range of institutions and WSI digital scanners, each containing substantial artefacts, and compare it to segmentations provided by Otsu thresholding and Histolab tissue segmentation and pen filtering tools. We found that our method segmented the tissue and fully removed all artefacts in 29 out of 30 WSIs, whereas Otsu thresholding failed to remove any artefacts, and the Histolab pen filtering tools only partially removed the pen marks. The beauty of our approach lies in its simplicity: manipulating RGB colour space and using Otsu thresholding allows for the segmentation of H&E-stained tissue and the rapid removal of artefacts without the need for machine learning or parameter tuning.

Stain normalization gives greater generalizability than stain jittering in neural network training for the classification of coeliac disease in duodenal biopsy whole slide images

Around 1% of the population of the UK and North America have a diagnosis of coeliac disease (CD), due to a damaging immune response to the small intestine. Assessing whether a patient has CD relies primarily on the examination of a duodenal biopsy, an unavoidably subjective process with poor inter-observer concordance. Wei et al. [11] developed a neural network-based method for diagnosing CD using a dataset of duodenal biopsy whole slide images (WSIs). As all training and validation data came from one source, there was no guarantee that their results would generalize to WSIs obtained from different scanners and laboratories. In this study, the effects of applying stain normalization and jittering to the training data were compared. We trained a deep neural network on 331 WSIs obtained with a Ventana scanner (WSIs; CD: n = 190 ; normal: n = 141 ) to classify presence of CD. In order to test the effects of stain processing when validating on WSIs scanned on varying scanners and from varying laboratories, the neural network was validated on 4 datasets: WSIs of slides scanned on a Ventana scanner (WSIs; CD: n = 48 ; normal: n = 35 ), WSIs of the same slides rescanned on a Hamamatsu scanner (WSIs; CD: n = 48 ; normal: n = 35 ), WSIs of the same slides rescanned on an Aperio scanner (WSIs; CD: n = 48 ; normal: n = 35 ), and WSIs of different slides scanned on an Aperio scanner (WSIs; CD: n = 38 ; normal: n = 37 ). Without stain processing, the F1 scores of the neural network were 0.947 , 0.619 , 0.746 , and 0.727 when validating on the Ventana validation WSIs, Hamamatsu and Aperio rescans of the Ventana validation WSIs, and Aperio WSIs from a different source respectively. With stain normalization, the performance of the neural network improved significantly with respective F1 scores 0.982 , 0.943 , 0.903 , and 0.847 . Stain jittering resulted in a better performance than stain normalization when validating on data from the same source F1 score 1.000 , but resulted in poorer performance than stain normalization when validating on WSIs from different scanners (F1 scores 0.939 , 0.814 , and 0.747 ). This study shows the importance of stain processing, in particular stain normalization, when training machine learning models on duodenal biopsy WSIs to ensure generalizability between different scanners and laboratories.

Multiple-instance-learning-based detection of coeliac disease in histological whole-slide images

We present a multiple-instance-learning-based scheme for detecting coeliac disease, an autoimmune disorder affecting the intestine, in histological whole-slide images (WSIs) of duodenal biopsies. We train our model to detect 2 distinct classes, normal tissue and coeliac disease, on the patch-level, and in turn leverage slide-level classifications. Using 5-fold cross-validation in a training set of 1841 (1163 normal; 680 coeliac disease) WSIs, our model classifies slides as normal with accuracy (96.7±0.6)%, precision (98.0±1.7)%, and recall (96.8±2.5)%, and as coeliac disease with accuracy (96.7±0.5)%, precision (94.9±3.7)%, and recall (96.5±2.9)% where the error bars are the cross-validation standard deviation. We apply our model to 2 test sets: one containing 191 WSIs (126 normal; 65 coeliac) from the same sources as the training data, and another from a completely independent source, containing 34 WSIs (17 normal; 17 coeliac), obtained with a scanner model not represented in the training data. Using the same-source test data, our model classifies slides as normal with accuracy 96.5%, precision 98.4% and recall 96.1%, and positive for coeliac disease with accuracy 96.5%, precision 93.5%, and recall 97.3%. Using the different-source test data the model classifies slides as normal with accuracy 94.1% (32/34), precision 89.5%, and recall 100%, and as positive for coeliac disease with accuracy 94.1%, precision 100%, and recall 88.2%. We discuss generalising our approach to screen for a range of pathologies.

Clinical standards for drug-susceptible pulmonary TB

BACKGROUND: The aim of these clinical standards is to provide guidance on 'best practice´ for diagnosis, treatment and management of drug-susceptible pulmonary TB (PTB).METHODS: A panel of 54 global experts in the field of TB care, public health, microbiology, and pharmacology were identified; 46 participated in a Delphi process. A 5-point Likert scale was used to score draft standards. The final document represents the broad consensus and was approved by all 46 participants.RESULTS: Seven clinical standards were defined: Standard 1, all patients (adult or child) who have symptoms and signs compatible with PTB should undergo investigations to reach a diagnosis; Standard 2, adequate bacteriological tests should be conducted to exclude drug-resistant TB; Standard 3, an appropriate regimen recommended by WHO and national guidelines for the treatment of PTB should be identified; Standard 4, health education and counselling should be provided for each patient starting treatment; Standard 5, treatment monitoring should be conducted to assess adherence, follow patient progress, identify and manage adverse events, and detect development of resistance; Standard 6, a recommended series of patient examinations should be performed at the end of treatment; Standard 7, necessary public health actions should be conducted for each patient. We also identified priorities for future research into PTB.CONCLUSION: These consensus-based clinical standards will help to improve patient care by guiding clinicians and programme managers in planning and implementation of locally appropriate measures for optimal person-centred treatment for PTB.

Risk factors for TB in Australia and their association with delayed treatment completion

BACKGROUND: Australia has a low incidence of TB and has committed to eliminating the disease. Identification of risk factors associated with TB is critical to achieving this goal.METHODS: We undertook a prospective cohort study involving persons receiving TB treatment in four Australian jurisdictions. Risk factors and their association with delayed treatment completion (treatment delayed by at least 1 month) were analysed using univariate analyses and multivariate logistic regression.RESULTS: Baseline surveys were completed for 402 persons with TB. Most (86.1%) were born overseas. Exposure to a person with TB was reported by 19.4%. Diabetes mellitus (10.2%), homelessness (9.2%), cigarette smoking (8.7%), excess alcohol consumption (6.0%) and mental illness (6.2%) were other common risk factors. At follow-up, 24.8% of patients had delayed treatment completion, which was associated with adverse events (34.1%, aOR 6.67, 95% CI 3.36-13.27), excess alcohol consumption (6.0%, aOR 21.94, 95% CI 6.03-79.85) and HIV co-infection (2.7%, aOR 8.10, 95% CI 1.16-56.60).CONCLUSIONS: We identified risk factors for TB and their association with delayed treatment completion, not all of which are routinely collected for surveillance purposes. Recognition of these risk factors should facilitate patient-centred care and assist Australia in reaching TB elimination.

Receiving healthcare for drug-resistant TB: a cross-sectional survey from Pakistan

OBJECTIVE

To describe and quantify patients' self-reported experiences of receiving healthcare from Pakistan's Programmatic Management of Drug-Resistant Tuberculosis (PMDT) model of care, and to understand these experiences within the broader context of Pakistan's health system.

METHOD

This was a cross-sectional survey of patients attending three PMDT clinics in Khyber-Pakhtunkhwa Province in Pakistan.

RESULTS

The median consultation time at the PMDT clinics was 10 minutes. In their most recent visit to the PMDT clinic, 34.9% of patients spent >40% of their monthly income to access treatment. To specify, 71% of patients reported spending out-of-pocket for ancillary medicines and 44.7% for laboratory tests. In 10.5% of cases, medicines for drug-resistant TB (DR-TB) were dispensed without the patient attending the clinic. Only 43.7% of treatment supporters regularly accompanied patients to the clinic, and 6% supervised the patient's intake of medicines. Disbursement of financial support was irregular in 98.6% of cases. Only 6.2% of patients received their daily injections from a public facility, the rest went elsewhere.

CONCLUSION

Several shortcomings in PMDT services, including hurried consultations, irregularities in financial support, and gaps in Pakistan's broader health system undermined healthcare experience of patients with DR-TB. To improve health outcomes and patients' care experience these service gaps need to be addressed.

WHO ethics guidance on TB care and migration: challenges to the implementation process

We summarise the current ethical guidance on tuberculosis (TB) care and migration, as set out in the WHO "Ethics Guidance for the Implementation of the End TB Strategy." Among other aspects, the Ethics Guidance states that there should be firm legal principles in place that ensure the enforcement of migration law on the one hand and the protection of human rights, including the right to health, on the other are separated from one another. As a challenge to the Ethics Guidance and its implementation, we describe two cases, each of which typifies particular problems. Case one describes the experience of a migrant worker in the United Arab Emirates who is deported when mandatory medical exams show evidence of current or prior TB. Case two raises the issue of providing more than TB care, which may also be needed for holistic care. The paper concludes with our suggestions for ways in which we could make progress towards ethically optimal TB care for migrants.

High-degeneracy Potts coarsening

I examine the fate of a kinetic Potts ferromagnet with a high ground-state degeneracy that undergoes a deep quench to zero temperature. I consider single spin-flip dynamics on triangular lattices of linear dimension 8≤L≤128 and set the number of spin states q equal to the number of lattice sites L×L. The ground state is the most abundant final state, and is reached with probability ≈0.71. Three-hexagon states occur with probability ≈0.26, and hexagonal tessellations with more than three clusters form with probabilities of O(10^{-3}) or less. Spanning stripe states-where the domain walls run along one of the three lattice directions-appear with probability ≈0.03. "Blinker" configurations, which contain perpetually flippable spins, also emerge, but with a probability that is vanishingly small with the system size.

Topology-controlled Potts coarsening

We uncover unusual topological features in the long-time relaxation of the q-state kinetic Potts ferromagnet on the triangular lattice that is instantaneously quenched to zero temperature from a zero-magnetization initial state. For q=3, the final state is either the ground state (frequency ≈0.75), a frozen three-hexagon state (frequency ≈0.16), a two-stripe state (frequency ≈0.09), or a three-stripe state (frequency <2×10^{-4}). Other final state topologies, such as states with more than three hexagons, occur with probability 10^{-5} or smaller, for q=3. The relaxation to the frozen three-hexagon state is governed by a time that scales as L^{2}lnL. We provide a heuristic argument for this anomalous scaling and present additional new features of Potts coarsening on the triangular lattice for q=3 and for q>3.

Multidrug-resistant tuberculosis in Australia, 1998-2012

OBJECTIVE

To describe the epidemiology and outcomes of multidrug-resistant tuberculosis (MDR-TB) diagnosed in Australia between 1998 and 2012.

DESIGN

A retrospective review was undertaken involving all patients with laboratory-confirmed MDR-TB notified in Australia between 1998 and 2012 inclusive. Demographic, clinical and laboratory features are described. Clinical outcomes were defined according to World Health Organization definitions of treatment success (cure and treatment completion), treatment failure, death, loss to follow-up (including transfer out), or not evaluated at treatment completion.

RESULTS

A total of 244 cases of MDR-TB were diagnosed in Australia during the study period, representing 1.4% of all TB cases notified. The majority were born outside Australia, including one third in Papua New Guinea. Of those with treatment outcome data available, treatment success was demonstrated in 81%. Treatment success was positively associated with use of a second-line injectable agent. Those born in Papua New Guinea were less likely to achieve treatment success.

CONCLUSION

MDR-TB is uncommon in Australia. The large number of cases born in Papua New Guinea, and the poorer outcomes in this cohort, represent challenges with cross-border management of MDR-TB in the Torres Strait. Australia has an ongoing role in the prevention and management of MDR-TB locally and in the region.

Tuberculosis in adolescents and young adults: epidemiology and treatment outcomes in the Western Cape

SETTING

Western Cape Province, South Africa.

OBJECTIVES

To characterise tuberculosis (TB) epidemiology, disease presentation and treatment outcomes among adolescents (age 10-19 years) and young adults (age 20-24 years) in the Western Cape.

DESIGN

A retrospective, cross-sectional review of routine patient-level data from the Electronic TB Register (ETR.Net) for 2013. Site of TB disease, human immunodeficiency virus (HIV) status and TB treatment outcomes were analysed by 5-year age groups (<5, 5-9, 10-14, 15-19, 20-24 and 25 years of age). TB notification rates were calculated using census data.

RESULTS

Adolescents and young adults comprised 18.0% of all new TB notifications in 2013. The notification rate was 141 TB cases/100 000 person-years (py) among 10-14 year olds, 418/100 000 py among 15-19 year olds and 627/100 000 py among 20-24 year olds. HIV prevalence among TB patients was 10.9% in 10-14 year olds, 8.8% in 15-19 year olds and 27.2% in 20-24 year olds. Older adolescents (age 15-19 years) and young adults (age 20-24 years) with HIV co-infection had poor treatment outcomes: 15.6% discontinued treatment prematurely and 4.0% died.

CONCLUSIONS

Young people in the Western Cape suffer a substantial burden of TB, and those with TB-HIV co-infection are at high risk of treatment discontinuation.

Gender differences in tuberculosis diagnosis, treatment and outcomes in Victoria, Australia, 2002–2015

BACKGROUND: Gender has a significant impact on tuberculosis (TB) diagnosis and outcomes in many settings worldwide. We explored gender differences in Victoria, Australia, a low-incidence setting.METHODS: Retrospective cohort study: 2002–2015. Gender was included as an independent variable in multivariate statistical analyses assessing TB management.RESULTS: There were 2655 (54.5%) males and 2212 (45.5%) females notified (male:female ratio = 1.2:1). Among cases with pulmonary involvement, males underwent a chest X-ray or CT scan (CXR) sooner (hazard ratio [HR] 1.2, 95%CI 1.04–1.31, P = 0.010), began treatment sooner after presentation (HR 1.2, 95%CI 1.08–1.34, P = 0.001) and were more likely to have a sputum smear sample performed (OR 1.3, 95%CI 1.01–1.55, P = 0.037). Male cases with extra-pulmonary TB sought health care sooner after symptom onset (HR 1.3, 95%CI 1.03–1.58, P = 0.024) and were more likely to have an abnormal CXR (OR 1.9, 95%CI 1.54–2.32, P < 0.001). Males were more likely to die before or during treatment (OR 1.5, 95%CI 1.06–2.11, P = 0.024).CONCLUSIONS: Women experienced small delays in management compared with men, with no obvious detriment to assessment results or treatment outcomes. Differences were consistent with the hypothesis that males manifest more severe disease at presentation, which could be related to a range of biological and social factors.

Molecular epidemiology of tuberculosis in Victoria, Australia, reveals low level of transmission

SETTING

Victoria, Australia.

OBJECTIVE

To measure the level of Mycobacterium tuberculosis transmission in Victoria.

DESIGN

Retrospective analysis of mycobacterial interspersed repetitive unit-variable number of tandem repeats (MIRU-VNTR) typing profiles from all first M. tuberculosis complex (MTC) isolates obtained from patients residing in Victoria from 2003 to 2010 was performed. State TB reference laboratory records were matched with Department of Health notification records to obtain further laboratory, demographic, contact investigation, clinical and treatment data. These data were used to assign patients to one of four categories: 1) no epidemiological link, 2) possible link, 3) likely transmission event and 4) laboratory contamination.

RESULTS

A total of 2377 MTC isolates were genotyped using 15-locus MIRU-VNTR. Of the 2298 M. tuberculosis isolates, 1029 (44.8%) had unique genotypic profiles and were considered epidemiologically unrelated, while 1269 (55.2%) isolates shared a profile with one or more other strains, defined as a genotypic cluster. Systematic investigation of all 268 genotypic clusters, including 24-locus MIRU-VNTR on selected isolates, led to a further 862 patients being classified as unrelated, bringing the total number of patients with no epidemiological links to 1891 (82.3%). Of the remaining patients, 294 (12.8%) were classified as having possible epidemiological links, 96 (4.2%) were classified as having known epidemiological links representing likely transmission events and 17 (0.7%) as the result of laboratory cross-contamination.

CONCLUSIONS

There is considerable genotypic diversity among Victorian MTC isolates, and the level of transmission is low.

Research ethics in The Union: an 8-year review of the Ethics Advisory Group

SETTING

The Ethics Advisory Group (EAG) of the International Union Against Tuberculosis and Lung Disease (The Union) was established in 2004 to provide ethical guidance and promote ethical standards within The Union, including reviews of proposed research projects associated with The Union.

OBJECTIVES

To describe research proposal reviews conducted by the EAG in the period 2005-2012 in terms of 1) annual numbers, 2) the Union departments in which the proposals originated, 3) study designs, 4) regions and countries where studies were to be conducted, 5) study topics, 6) problems encountered by the EAG, and 7) review outcomes.

DESIGN

Descriptive study of application records of the EAG.

RESULTS

A total of 292 applications were reviewed; 79% were proposals for operational research; 85% were from Africa and Asia, with 64% from India, South Africa, Malawi, Kenya and Zimbabwe. Tuberculosis was the topic in 68%; only three studies in the 8 years were on other lung diseases. Several problems encountered are highlighted. All applications were approved except six, either immediately or after modification.

CONCLUSION

The proposal review process of the EAG serves to maintain ethical standards of research within The Union. Ideas for expanding the scope of the EAG are discussed.

Triaging pandemic flu: pneumonia severity scores are not the answer

SETTING

The 2009 H1N1 influenza pandemic caused significant strain on health systems worldwide. A tool to triage patients at low risk of requiring intensive care services would assist practitioners in safely reducing hospital admission rates during pandemic influenza outbreaks. Community-acquired pneumonia severity scores have not been validated for use in pandemic influenza.

OBJECTIVE

To assess the accuracy of the pneumonia severity index (PSI), CURB-65 and SMRT-CO severity scores in predicting patients at low risk of requiring intensive care services.

DESIGN

Between May and July 2009, 105 patients admitted with laboratory-confirmed pandemic (H1N1) 2009 influenza to Melbourne public hospitals were assessed on admission to determine their pneumonia severity scores and subsequent need for intensive care unit (ICU) support and length of stay.

RESULTS

SMRT-CO was the most accurate score at predicting ICU admission, with an area under the curve of the receiver operating characteristic of 0.826. No score provided good discrimination of low-risk patients, with respectively 19%, 21% and 12% requiring ICU admission as predicted by PSI, CURB-65 and SMRT-CO.

CONCLUSION

Current pneumonia severity scores have insufficient predictive ability to safely identify low-risk patients with pandemic (H1N1) 2009 influenza.