Delivering robust and meaningful data in support of an ever-increasing demand on tissue-based biomarker evaluations is critical to improving decision making in clinical R&D. To realise these requirements, it is vitally important to implement quality standards within image analysis workflows, to provide verified evidence supporting the value of data as part of tissue-based quantitative studies.

In this blog we highlight some of the key quality components required within an image analysis facility, including establishing a “Quality Management Framework”, defining ‘Quality Criteria’ and implementing ‘Quality Systems & Process Validations’.

Establish a Quality Management Framework

In order to manage, monitor and continuously improve quality as a function, it is necessary to establish a ‘Quality Management Framework’ that includes control procedures to support the verification of data production from images and assurance of associated activities and processes aligned to data production. Such a framework will enable data generated from clinical studies to be fully traceable and auditable by key stakeholders. As part of the process, it is also essential to ensure Quality Assurance, utilising external consultants   to perform independent auditing of facilities and procedures.

Define Quality Criteria

In order to ensure ‘quality data’ it is necessary to operate a range of ‘Quality Criteria’ including associated controls within image analysis workflow processes. These practises involve the application of knowledge and expertise regarding 5 key impact factors on quantitative data outputs:

• Study design – Optimal study design to provide quantitative analysis end-points associated with scientific rationale and study objective
• Technical qualities -Identification of process-related artefacts and associated consequences on quantitative outputs arising from tissue/section preparation
• Image parameters – Operation of digital imaging platforms and influence of scan parameters associated with image generation
• Algorithm verification – Development and implementation of image analysis algorithms
• Data significance – Interrogation and interpretation of quantitative analysis data

Validate the Systems and Processes

Quality systems validation involve 3 main qualification steps: i) Installation (IQ), ii) Operation (OQ) and iii) Performance (PQ) to support and ensure the validity and integrity of image analysis data produced therein. In addition, it is also important to apply study management and monitoring with continuous control and verification processes throughout the data production lifecycle including:

• Protocol design – Proactive working with key stakeholders to define protocols specifically designed towards image analysis endpoints
• Pathology communications – Working directly with Pathologists (ideally using an integrated cloud-based system) to generate annotations, assist with tumour/tissue identification and agreements on algorithm threshold optimisations prior to analysis.
• Assay validation support – Employing quantitative analysis on antibody dilution and precision studies to assist in the verification of IHC/IF assay sensitivity and specificity
• Chromogen sequence validations – Operating a series of quantitative test combinations on IHC chromogenic sequences to determine potential spectral masking effects on complex colour deconvolution processes and ensure consistent colour separations
• Image quality – Applying a six-point visual quality control practise to assess the potential impact of random factors, associated with slide scanning, tissue/section preparation, stain dynamics and exogenous artefacts, prior to image analysis.
• Algorithm validation – Utilising a comprehensive algorithm verification process to determine suitability, functionality and reliability in association with performance against a ‘gold standard’ or defined set of reference samples
• Data verification – Data reviews are performed throughout manual and automated image analysis phases to verify system parameters applied and data inputs, outputs and reports.

Implementation of the above quality control and assurance processes facilitate the delivery of high quality analysis and data in support of tissue-based quantitative requirements on clinical R&D studies. Ultimately, these processes help to enhance data quality and provide confidence in decision making.

More about the author: Alison Bigley is part of the OracleBio management team and previously worked at AstraZeneca where she was instrumental in implementing histology image analysis quality management processes across the Global Safety Pathology network.