How much reproducibility do we need in human and veterinary pathology?

Pospischil, Andreas; Folkers, Gerd

doi:10.1016/j.etp.2014.11.005

Cited by 3 publications

(1 citation statement)

References 24 publications

(9 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Providing more time to perform the corrections may also reduce the possible number of user errors in the object tracking, relabeling or trajectory editing processes, thereby raising efficiency and repeatability. Repeatability (or reproducibility) describes the certainty by which different people under varying laboratory conditions can reproduce all the procedures and methodologies from start to finish, thereby describing the accuracy and reliability of a test method [124]. These criteria closely match the criteria outlined in the ISO/IEC standards for usability metrics, which are outlined below.…”

Section: Evaluation Methodologymentioning

confidence: 88%

KOLAM : human computer interfaces fro visual analytics in big data imagery

Haridas¹

View full text Add to dashboard Cite

In the present day, we are faced with a deluge of disparate and dynamic information from multiple heterogeneous sources. Among these are the big data imagery datasets that are rapidly being generated via mature acquisition methods in the geospatial, surveillance (specifically, Wide Area Motion Imagery or WAMI) and biomedical domains. The need to interactively visualize these imagery datasets by using multiple types of views (as needed) into the data is common to these domains. Furthermore, researchers in each domain have additional needs: users of WAMI datasets also need to interactively track objects of interest using algorithms of their choice, visualize the resulting object trajectories and interactively edit these results as needed. While software tools that fulfill each of these requirements individually are available and well-used at present, there is still a need for tools that can combine the desired aspects of visualization, human computer interaction (HCI), data analysis, data management, and (geo-)spatial and temporal data processing into a single flexible and extensible system. KOLAM is an open, cross-platform, interoperable, scalable and extensible framework for visualization and analysis that we have developed to fulfil the above needs. The novel contributions in this thesis are the following: 1) Spatio-temporal caching for animating both giga-pixel and Full Motion Video (FMV) imagery, 2) Human computer interfaces purposefully designed to accommodate big data visualization, 3) Human-in-the-loop interactive video object tracking - ground-truthing of moving objects in wide area imagery using algorithm assisted human-in-the-loop coupled tracking, 4) Coordinated visualization using stacked layers, side-by-side layers/video sub-windows and embedded imagery, 5) Efficient one-click manual tracking, editing and data management of trajectories, 6) Efficient labeling of image segmentation regions and passing these results to desired modules, 7) Visualization of image processing results generated by non-interactive operators using layers, 8) Extension of interactive imagery and trajectory visualization to multi-monitor wall display environments, 9) Geospatial applications: Providing rapid roam, zoom and hyper-jump spatial operations, interactive blending, colormap and histogram enhancement, spherical projection and terrain maps, 10) Biomedical applications: Visualization and target tracking of cell motility in time-lapse cell imagery, collecting ground-truth from experts on whole-slide imagery (WSI) for developing histopathology analytic algorithms and computer-aided diagnosis for cancer grading, and easy-to-use tissue annotation features.

show abstract

Section: Evaluation Methodologymentioning

confidence: 88%

KOLAM : human computer interfaces fro visual analytics in big data imagery

Haridas¹

View full text Add to dashboard Cite

show abstract

Swiss Canine Cancer Registry 1955–2008: Occurrence of the Most Common Tumour Diagnoses and Influence of Age, Breed, Body Size, Sex and Neutering Status on Tumour Development

Grüntzig

Graf

Boo

et al. 2016

Journal of Comparative Pathology

120

View full text Add to dashboard Cite

This study is based on the Swiss Canine Cancer Registry, comprising 121,963 diagnostic records of dogs compiled between 1955 and 2008, in which 63,214 (51.83%) animals were diagnosed with tumour lesions through microscopical investigation. Adenoma/adenocarcinoma (n = 12,293, 18.09%) was the most frequent tumour diagnosis. Other common tumour diagnoses were: mast cell tumour (n = 4,415, 6.50%), lymphoma (n = 2,955, 4.35%), melanocytic tumours (n = 2,466, 3.63%), fibroma/fibrosarcoma (n = 2,309, 3.40%), haemangioma/haemangiosarcoma (n = 1,904, 2.80%), squamous cell carcinoma (n = 1,324, 1.95%) and osteoma/osteosarcoma (n = 842, 1.24%). The relative occurrence over time and the most common body locations of those tumour diagnoses are presented. Analyses of the influence of age, breed, body size, sex and neutering status on tumour development were carried out using multiple logistic regression. In certain breeds/breed categories the odds ratios (ORs) for particular tumours were outstandingly high: the boxer had higher ORs for mast cell tumour and haemangioma/haemangiosarcoma, as did the shepherd group for haemangioma/haemangiosarcoma, the schnauzer for squamous cell carcinoma and the rottweiler for osteoma/osteosarcoma. In small dogs, the risk of developing mammary tumours was three times higher than in large dogs. However, small dogs were less likely to be affected by many other tumour types (e.g. tumours of the skeletal system). Examination of the influence of sex and neutering status on tumour prevalence showed that the results depend on the examination method. In all sampling groups the risk for female dogs of developing adenoma/adenocarcinoma was higher than for male dogs. Females had a lower risk of developing haemangioma/haemangiosarcoma and squamous cell carcinoma than males. Neutered animals were at higher risk of developing specific tumours outside the genital organs than intact animals. The sample size allows detailed insight into the influences of age, breed, body size, sex and neutering status on canine tumour development. In many cases, the analysis confirms the findings of other authors. In some cases, the results are unique or contradict other studies, implying that further investigations are necessary.

show abstract

Inter-Rater and Intra-Rater Agreement in Scoring Severity of Rodent Cardiomyopathy and Relation to Artificial Intelligence–Based Scoring

Steinbach,

Tokarz,

et al. 2024

Toxicol Pathol

View full text Add to dashboard Cite

We previously developed a computer-assisted image analysis algorithm to detect and quantify the microscopic features of rodent progressive cardiomyopathy (PCM) in rat heart histologic sections and validated the results with a panel of five veterinary toxicologic pathologists using a multinomial logistic model. In this study, we assessed both the inter-rater and intra-rater agreement of the pathologists and compared pathologists’ ratings to the artificial intelligence (AI)-predicted scores. Pathologists and the AI algorithm were presented with 500 slides of rodent heart. They quantified the amount of cardiomyopathy in each slide. A total of 200 of these slides were novel to this study, whereas 100 slides were intentionally selected for repetition from the previous study. After a washout period of more than six months, the repeated slides were examined to assess intra-rater agreement among pathologists. We found the intra-rater agreement to be substantial, with weighted Cohen’s kappa values ranging from k = 0.64 to 0.80. Intra-rater variability is not a concern for the deterministic AI. The inter-rater agreement across pathologists was moderate (Cohen’s kappa k = 0.56). These results demonstrate the utility of AI algorithms as a tool for pathologists to increase sensitivity and specificity for the histopathologic assessment of the heart in toxicology studies.

show abstract

How much reproducibility do we need in human and veterinary pathology?

Cited by 3 publications

References 24 publications

KOLAM : human computer interfaces fro visual analytics in big data imagery

KOLAM : human computer interfaces fro visual analytics in big data imagery

Swiss Canine Cancer Registry 1955–2008: Occurrence of the Most Common Tumour Diagnoses and Influence of Age, Breed, Body Size, Sex and Neutering Status on Tumour Development

Inter-Rater and Intra-Rater Agreement in Scoring Severity of Rodent Cardiomyopathy and Relation to Artificial Intelligence–Based Scoring

Contact Info

Product

Resources

About