Šeškinės Poliklinika, one of the largest public clinics in Lithuania, provides care for more than 85,000 patients and manages thousands of imaging procedures every month. Facing increasing demand, an evolving clinical landscape, and the constant pressure to deliver high-quality care within limited time and resources, the radiology department began exploring how artificial intelligence (AI) could support their diagnostic workflows.
Over the past four years, the department has implemented the Oxipit CXR Suite – a portfolio of AI tools tailored specifically for chest X-rays. These tools have helped the clinic transition from experimentation to practical, measurable improvements in daily operations, creating a safer, faster, and more collaborative diagnostic environment. In this case study, Dr. Antanas Pempe, Head of the Radiology Department, shares the clinic’s AI adoption journey, highlighting both challenges and clinical benefits.
From curiosity to clinical integration
At the time of adoption, AI in radiology was still in its early stages. The decision to introduce Oxipit’s technology was not driven by a backlog crisis but rather by a progressive mindset – an interest in exploring tools that could improve consistency, efficiency, and patient outcomes. While leadership recognized the potential, initial clinical enthusiasm was mixed.
“There was some resistance,” Dr. Pempe notes. “Radiologists are trained to trust their own judgment. Some said, ‘We don’t need AI – we’re already doing a great job.’”
The initial implementation was also technically limited, with AI operating outside the PACS system. This required additional steps for radiologists to access AI outputs, making it less likely to be adopted in routine practice. As integration improved and the tools became part of the native workflow, engagement began to grow. The department began shifting focus from evaluating the concept to realizing value.
Putting AI into practice: seamless quality assurance and autonomous reporting
Today, the clinic uses Oxipit’s AI in two primary capacities: real-time quality assurance and autonomous healthy CXR reporting. Both are fully integrated into the PACS, supporting radiologists and referring physicians within the tools they already use. This seamless access to AI-generated insights has been essential for adoption and sustained use.
Oxipit’s quality assurance solution continuously reviews chest X-ray reports and flags cases where AI findings do not align with radiologist conclusions. These alerts trigger a second review, not as a judgment on performance, but as an opportunity to verify and ensure diagnostic consistency.
The Head of Department emphasizes a collaborative approach. “When there’s a discrepancy, I review the case with the reporting radiologist. We discuss it together – not to assign blame, but to learn and make sure the patient receives the best care.”
This layer of AI oversight acts as a silent safety net. While disagreements are rare – typically one message every 30 to 60 days – the ability to catch subtle findings or ambiguous cases contributes to overall diagnostic confidence and continuous improvement.
In addition to quality checks, the clinic has also implemented Oxipit CXR Suite’s autonomous reporting tool, ChestLink, for prophylactic (occupational) health checks. These exams are often required for job applications, licensing, or periodic screening, and typically involve healthy individuals.
ChestLink automatically clears chest X-rays that show no abnormalities with a high degree of certainty. Radiologists are only required to review studies where the AI detects even minimal deviations. This setup, designed with safety in mind, currently allows AI to autonomously report approximately 80% of these cases – significantly reducing routine workload.
This is where we see the most immediate impact. Patients don’t have to wait for radiologist confirmation. The examining physician can proceed immediately, knowing the scan was normal.
“This is where we see the most immediate impact,” the department head explains. “Patients don’t have to wait for radiologist confirmation. The examining physician can proceed immediately, knowing the scan was normal.”
By removing the bottleneck of manual review for healthy cases, the department accelerates service delivery without compromising care. The remaining cases still receive full radiologist attention, ensuring clinical oversight where it matters most.
Clinical impact and outcomes
- Faster decision-making at point of care
The benefit of AI is not confined to the radiology department. Referring physicians – including family doctors and pediatricians – now have immediate access to AI impressions in the PACS. This allows them to make quicker clinical decisions, especially in urgent or high-volume situations.
“During flu season or COVID outbreaks, we often see queues of patients. AI-generated impressions help doctors decide on next steps without waiting for a full report,” the department head explains. “It shortens the loop between imaging and treatment.”
The immediacy of AI results helps avoid delays, improves triage, and ultimately enhances patient experience. - Reduced radiologist workload
With over 2,700 chest X-rays processed each month, even marginal gains translate into substantial time savings. By shifting the burden of healthy case review to AI, radiologists can dedicate more time to complex cases and consultations.
“We don’t have advanced modalities like PET-CT or cardiac MRI. Chest X-rays are our core workload,” Dr. Pempe says. “AI helps us focus our attention on findings that actually require human interpretation.”
This division of labor has created a more balanced and sustainable work environment. - Building radiologist trust through transparent, collaborative use
Initial skepticism has faded, particularly as the department took a collaborative and transparent approach to AI. Case discussions involving AI flags are now part of regular quality assurance, and many radiologists view the AI system as a useful second set of eyes.
“When I receive a notification from the AI, I check who reported the case and we discuss it. It’s become a routine part of our checks,” says the department lead.
Younger radiologists, in particular, welcome the automation of routine tasks. One team member described healthy CXRs as offering “no educational value,” preferring to spend time on diagnostic challenges instead. This shift in perception reflects a growing understanding of AI’s role as an enabler, not a threat. - Future readiness
With the Lithuanian Ministry of Health preparing to expand screening programs – including low-dose CT for lung cancer and extended mammography eligibility – the department anticipates a sharp rise in imaging volume.
“We’re already seeing demand rise. If we don’t use AI, we’ll be overwhelmed,” the department head says. “This is not about luxury or experimentation – it’s about readiness.”
Having AI systems in place now means the clinic can absorb these future demands without compromising care or overstretching staff.
Cultural shift
The most meaningful change may be the shift in mindset. Through deliberate leadership, transparent discussions, and consistent use, AI has gone from a theoretical tool to a trusted part of clinical care.
“Radiologists take pride in their work – and rightly so. But they also recognize that collaboration, whether with colleagues or AI, leads to better care,” says Dr. Pempe.
Integration into PACS systems and clinical routines was key. “Once it was seamless, the resistance faded. Now, it’s just part of how we work.”
Radiologists take pride in their work – and rightly so. But they also recognize that collaboration, whether with colleagues or AI, leads to better care.
What’s next: measuring the impact and planning for full automation
The clinic is now exploring how to more formally quantify the benefits – including time savings, patient wait time reductions, and workload shifts. While reimbursement models still require radiologists to sign off on all reports, future legal changes could enable full automation for select workflows.
“Doctors are already saving time. Patients are being seen faster. We know it works. Now we want to measure it more clearly,” says the department head.
Advice to peers: start now or risk falling behind
When asked what advice he would give peers considering AI, the response was immediate:
“If you haven’t done it yet – do it now. AI is already part of our daily lives. If you wait, you’ll fall behind.”
With clinical, operational, and future-readiness benefits already in place, Šeškinės Poliklinika shows how AI can be quietly transformative when introduced thoughtfully and used consistently.