Add blog post for image reading

app/posts/manage-breast-screening/2025/04/2025-04-08-understanding-image-reading.md

@@ -0,0 +1,80 @@
+---
+  title: Understanding image reading
+  description: How image reading currently works in breast screening
+  date: 2025-04-08
+  tags:
+    - alpha
+    - prototype
+    - image reading
+---
+
+## What is image reading
+
+Image reading (or film reading) is the process where specially trained healthcare professionals review mammograms taken during breast screening, looking for abnormalities. It's colloquially called 'film reading' because traditionally they would be reviewing physical x-ray films. Although mammograms are now digital, the term 'film reading' remains in common use.
+
+## How image reading works
+
+Mammogram images are primarily read by consultant radiologists but can also be read by advanced practitioner radiographers and breast clinicians who have undergone specific training in mammogram interpretation. For consistency and safety, each participant's set of mammograms is read twice (called first read and second read) by two different readers.
+
+Each mammogram typically consists of four standard views - right and left breast, each usually with two different angles (craniocaudal and mediolateral oblique views). The reader examines all views carefully, looking for subtle signs that might indicate cancer. If available they’ll compare these to prior mammograms to look for any changes.
+
+The exact process varies by breast screening unit, but the general workflow is:
+
+1. **Prepare clinic**: Admin staff prepare paperwork and check all cases are ready for reading
+2. **First read**: A reader reviews the images and records their opinion
+3. **Second read**: A different reader examines the same images
+4. **Results comparison**: The system compares the two readings
+
+Many units use 'blind reading' where the second reader doesn't see the first reader's opinion, but this practice isn't universal across all screening units.
+
+If both readers agree the images do not show any abnormalities, then the participant is classed as 'routine recall' and will be told their results are normal. This is the most common result and happens about 90% of the time.
+
+If the the two readers disagree, then the case will go to 'arbitration' or 'consensus' - where further discussion by a third reader or panel takes place to decide whether to recall for assessment or give a normal result.
+
+If both agree abnormalities are present, then the participant will be asked back to the breast screening unit for further tests. This is known as ‘assessment’. Some units will take these to arbitration first.
+
+Two additional outcomes can occur:
+
+* **Technical recall**: If image quality issues prevent proper interpretation, the participant is asked back for new mammograms
+* **Clinical recall**: If the participant reported significant symptoms during screening, they may be recalled for assessment regardless of the mammogram findings
+
+Readings get recorded in National Breast Screening System (NBSS) where some minimal other data can be captured. Readers may also record other clinical data on paper or in clinic or department logs.
+
+## Why we're looking at it now
+
+Image reading is a key stage in the breast screening process, but is heavily paper-based in many units, relying on shuffling groups of paper forms around with lots of manual cross-checking to ensure no one is forgotten. When recording outcomes, the NBSS is limited in the detail that can be captured - with potential for other details to be recorded on paper again.
+
+Image reading represents one of the main stages that would need to modernise to support a fully digital breast screening service.
+
+## How reading works day-to-day
+
+For routine screening, image reading happens using NBSS – but is still heavily paper-based. Admins staff carefully look after this process, making sure everyone who was screened in a clinic is ready for reading.
+
+If screening identified that the participant had had mammograms at another location previously, admin staff may attempt to retrieve copies of these before putting that case to reading - this is known as ‘awaiting priors’. Not all units do this - some will only request prior mammograms after reading if concerns are seen.
+
+Once ready for reading, the bundles of screening forms grouped by clinic are usually kept on shelves in date order. In NBSS, you can select a clinic to read in date order, but in practice breast screening units will use the physical stacks on the shelves. These stacks may be grouped by clinics that have not been read yet (need a first read) and those that have had one read and need a second read.
+
+A reader will pick a bundle from the shelf – usually the oldest – and take it to their reading desk. They'll then open the corresponding clinic in NBSS. Working through the stack of paper one participant at a time, NBSS will cycle through the same participants in the same order.
+NBSS is used to record the outcome of the reading, but not to display the mammograms. When a participant is opened in NBSS, the reader's PACS (Picture Archiving and Communication System) workstation will automatically open the corresponding mammograms and any prior images for comparison. The reader will review the mammograms, then input their opinion in NBSS.
+
+The process includes a fair amount of cross-checking with the paper forms – verifying that the count of sheets in the bundle matches the number read on NBSS, signing each paper form, reading annotations from the paper forms, and perhaps making further annotations on the paper forms.
+
+## Challenges with the current process
+
+The current approach presents several challenges:
+
+* Time-consuming manual processes that could be automated
+* Risk of human error in cross-checking and form handling
+* Rigid clinic-based organisation that may not always be the most efficient workflow, or limits exploring other workflows
+* Limited capture of detailed reading information that could be valuable for quality assurance
+
+## What a digital solution could offer
+
+A modern digital approach to image reading could:
+
+* Streamline the workflow by eliminating paper handling
+* Allow for more flexible reading workflows beyond the traditional clinic-based approach
+* Capture more detailed information about findings
+* Improve audit trails and quality assurance
+* Support training and continued professional development through better feedback
+