{"id":447194,"date":"2026-05-27T04:49:16","date_gmt":"2026-05-27T04:49:16","guid":{"rendered":"https:\/\/www.newsbeep.com\/nz\/447194\/"},"modified":"2026-05-27T04:49:16","modified_gmt":"2026-05-27T04:49:16","slug":"ai-may-soon-help-read-mammograms-in-nz-expert-reaction","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/nz\/447194\/","title":{"rendered":"AI May Soon Help Read Mammograms In NZ \u2013 Expert Reaction"},"content":{"rendered":"

\"Science<\/a>
\nWednesday, 27 May 2026, 3:34 pm
Press Release: Science Media Centre<\/a><\/p>\n

The government is choosing a mammogram-reading AI tool to
\ntest and validate, ahead of a planned roll out early next
\nyear.<\/p>\n

The tool will ease clinician workloads by doing
\none of the two independent reads currently required during
\nmammogram assessment, the government says. They emphasise
\nthat all diagnoses and follow-up decisions will continue to
\ninclude qualified health professionals.<\/p>\n

The Science
\nMedia Centre asked experts to comment.<\/p>\n

Associate
\nProfessor Catherine Shi, Head of the Department of Data
\nScience & AI, AUT, comments:<\/p>\n

\u201cThis is a very
\npositive step for New Zealand and shows increasing maturity
\nin translating AI research into real healthcare
\napplications. Breast screening programmes generate extremely
\nhigh workloads for radiologists, and AI tools like this have
\nstrong potential to improve screening efficiency and help
\ndetect subtle abnormalities that may otherwise be
\nmissed.<\/p>\n

\u201cIt is particularly encouraging to see the
\nproject being developed alongside healthcare professionals
\nand integrated into existing clinical workflows, because
\nsuccessful healthcare AI deployment depends not only on
\nmodel accuracy, but also on how clinicians interpret, trust,
\nand use the system safely in everyday
\npractice.\u201d<\/p>\n

Conflict of interest statement: \u201cI
\ndeclare no real or potential conflict of interest with this
\nproject.\u201d<\/p>\n

Dr Nicholas Knowlton, Senior Lecturer
\nin Statistics, Massey University, comments:<\/p>\n

\u201cThe
\nuse of AI as a second reader in mammography is timely and,
\nin my view, relatively low risk when implemented properly.
\nThe key point is that this is not AI replacing radiologists
\nor making treatment decisions. It is a decision-support
\ntool. Like spell check, it can flag what it thinks it sees,
\nbut a human still must use clinical
\njudgement.<\/p>\n

Advertisement – scroll to continue reading<\/p>\n

\u201cRadiologists already use software tools
\nto manipulate images, improve contrast, and help visualise
\nstructures in the breast. AI adds another layer by
\nindicating which image features may be suspicious or worth
\ncloser attention. The final interpretation still sits with
\nthe radiologist. Patients are still being assessed by a
\ndoctor.<\/p>\n

\u201cThe more important risk is not the tool
\nitself, but the governance of the data behind it. New
\nZealand has high-value public mammography data. If that data
\nare allowed to leave the country and contribute to the
\ndevelopment or improvement of commercial AI systems
\noverseas, while New Zealand researchers cannot access
\ncomparable data to build local capability, then we are
\neffectively exporting public value.<\/p>\n

\u201cThese datasets
\nare not just technical assets. In Aotearoa, they include
\ninformation from our communities and should be treated as
\ntaonga. There needs to be a fairer national approach that
\nenables safe, privacy-preserving access for New Zealand
\nresearchers, so the benefits of AI in breast screening are
\nnot captured mainly offshore.<\/p>\n

\u201cThe clinical
\nimplementation may be sensible. The data governance question
\nis the one we should not ignore.\u201d<\/p>\n

Conflict of
\ninterest statement: \u201cI work commercially in the
\nreproductive space in addition to my academic work, so
\nshouldn\u2019t have any conflicts with this
\nstatement.\u201d<\/p>\n

Ross Lawrenson, Professor of
\nPopulation Health, University of Waikato,
\ncomments:<\/p>\n

\u201cIt is good to see the Minister announce
\ninvestment in the use of AI to improve the efficiency and
\neffectiveness of our breast screening programme. New Zealand
\nhas had a national mammography screening programme since
\n1998, and there has always been a challenge in balancing the
\nsensitivity and specificity of screening. In other words, we
\ndo not want to miss cases of breast cancer, while at the
\nsame time we want to avoid incorrectly labelling women as
\nhaving cancer when they do not. No screening system is 100%
\nsensitive and specific.<\/p>\n

\u201cWhen screening relies on
\nindividual clinicians interpreting mammography images, we
\nknow there is considerable variability in identifying
\nsuspicious breast lumps that require biopsy. False-positive
\nrates, for example, can range from roughly 3% to more than
\n15%. To reduce this variability, BreastScreen Aotearoa
\nmandates double reporting of all mammograms. While this
\nimproves consistency and diagnostic accuracy, it also
\nincreases workforce demands and
\ncosts.<\/p>\n

\u201cComputer-assisted diagnosis (CAD) has been
\nused to support mammogram interpretation for several
\ndecades. More recently, the rapid expansion of AI
\ntechnologies in mammography has generated interest in their
\npotential to improve both the sensitivity and specificity of
\nthe current double-reading system used by expert
\nradiologists.<\/p>\n

\u201cAI tools can assign risk scores to
\nbreast lumps identified on mammography, categorising them as
\nhigher or lower risk. These scores can assist radiologists
\nin clinical decision-making, including determining whether
\nfollow-up imaging or biopsy is required. Current evidence
\nsuggests that the improvements achieved with AI are modest
\nbut meaningful. For example, one large study found that
\nimplementing AI-assisted additional reading could detect an
\nadditional 0.7\u20131.6 cancers per 1000 women
\nscreened.<\/p>\n

\u201cHowever, AI systems are less accurate in
\nwomen with dense breast tissue. This highlights the
\ncontinuing importance of routinely recording breast density
\nand ensuring additional support and surveillance for
\nhigher-risk women with dense breasts \u2014 an area that
\nremains a challenge for our screening system. There is
\ngrowing agreement that, particularly for lower-risk women,
\nAI-assisted diagnostics may safely replace the need for
\ndouble reading, helping to reduce pressure on radiology
\nservices while maintaining screening
\nquality.\u201d<\/p>\n

Conflict of interest statement: \u201cNo
\nconflicts of
\ninterest.\u201d<\/p>\n

\n\u00a9 Scoop Media<\/a><\/p>\n

\"Science<\/a><\/p>\n

Our aim is to promote accurate, evidence-based reporting on science and technology by helping the media work more closely with the scientific community.<\/p>\n

The Science Media Centre is New Zealand’s only trusted, independent source of information for the media on all issues related to science. Thousands of news stories providing context from and quoting New Zealand researchers have been published as a direct result of our work.<\/p>\n

\u00a0<\/p>\n","protected":false},"excerpt":{"rendered":"Wednesday, 27 May 2026, 3:34 pmPress Release: Science Media Centre The government is choosing a mammogram-reading AI tool…\n","protected":false},"author":2,"featured_media":35750,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[10],"tags":[365,134,227957,2844,111,24636,139,69,147673,32292,39761,147,227956,11112,145],"class_list":["post-447194","post","type-post","status-publish","format-standard","has-post-thumbnail","category-health","tag-ai","tag-health","tag-healthcare-applications","tag-mammograms","tag-new-zealand","tag-newsworthy","tag-newzealand","tag-nz","tag-primary-health","tag-radiology","tag-sci-tech","tag-science","tag-science-media-centre","tag-scoop","tag-technology"],"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/posts\/447194","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/comments?post=447194"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/posts\/447194\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/media\/35750"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/media?parent=447194"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/categories?post=447194"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/tags?post=447194"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}