CONNECT WITH US
Friday 6 March 2026
How NeuroSpine AI Is Rewriting Surgical Planning
When spine surgeons talk about pressure, few procedures rival C1–C2 fixation - a surgery so delicate that a deviation of just a few millimeters can mean the difference between success and catastrophic complications.That reality became the starting point for NeuroSpine AI, a project developed by Sanny Kumar Sahani, a PhD researcher in computer science and commercial engineering, and Shweta Prajapati, a master’s student in biomedical engineering. Both are from India, both study at Chang Gung University, and both work in the same lab under the same advisor.Their collaboration has earned them international recognition- the Bronze Medal at Taiwan's Ministry of Economic Affairs Best AI Award, standing out in a field crowded with enterprise and academic innovations.Sahani recalls that during collaboration with Chang Gung Memorial Hospital, one experienced spine surgeon, Dr. Wu, described the mental burden of C1–C2 screw planning. Even for veteran surgeons, the risk remains high. The anatomy is complex, patient variation is significant, and the vertebrae sit dangerously close to the brainstem, nerves, and major blood vessels. If automation and precision could be applied anywhere in spine surgery, this was it.Prajapati emphasizes that the motivation was never purely academic. Surgeons identified real constraints: planning is time-consuming, only highly experienced specialists can safely perform these procedures, and screw misplacement remains a serious clinical risk. The team's goal became clear - reduce planning time, lower dependence on elite expertise, and improve consistency without compromising safety.Most existing surgical planning tools focus on the thoracic and lumbar spine. Those vertebrae are relatively uniform, making automation easier. C1 and C2, by contrast, are anatomically unique, highly variable, and structurally complex.Sahani explains that current solutions either require extensive manual adjustment or do not support C1–C2 at all. NeuroSpine AI took the opposite approach: start with the hardest problem first.The system automatically generates multiple screw trajectories, performs geometric safety checks, and ensures consistency regardless of who uses the software. Unlike manual planning - which varies between surgeons and depends heavily on experience - AI-generated trajectories remain stable and repeatable.That consistency matters, especially for junior surgeons still building experience. The system does not replace surgical judgment, but it removes unnecessary variability from one of the most critical planning steps.In traditional workflows, planning a single screw trajectory can take 30 to 60 minutes. NeuroSpine AI generates multiple trajectories in just two to three minutes.More importantly, the system has been trained to understand the geometric patterns of C1–C2 anatomy, learning from diverse datasets collected through Chang Gung Memorial Hospital and an international collaboration in France. The AI does not simply segment images; its reason about spatial constraints, vessel proximity, and safe paths for screw placement.The result is a pre-operative planning tool that balances speed, safety, and accuracy — something surgeons rarely get at the same time.The role of AI is more Than just automation. First, it specializes exclusively in C1–C2 anatomy, rather than applying generalized spine models. Second, it performs geometric reasoning to avoid nerve and vessel damage - a non-negotiable requirement given the proximity to the brain. Third, it generates multiple alternative trajectories, ensuring that even abnormal anatomical cases still yield viable surgical options.Prajapati notes that the system is designed for pre-operative use. Surgeons can review trajectories before entering the operating room, making the procedure safer and more predictable - particularly for less experienced doctors.NeuroSpine AI has already completed initial clinical validation at Chang Gung Memorial Hospital, where surgeons confirmed that AI-generated trajectories aligned closely with what experienced clinicians would plan manually.That validation marked a turning point. The project is now transitioning from research to a deployable product.The next phase involves expanding beyond C1–C2 to cover the entire spine - all 26 vertebrae - and integrating the system into existing clinical workflow software. Given that C1–C2 is the most complex region, the team believes scaling to other vertebrae is both realistic and strategic.The potential market spans hospitals, medtech companies, surgical planning platforms, robotic surgery firms, and spine implant manufacturers. As spine surgeries increase globally, automated pre-operative planning is becoming less optional and more essential.The team plans to begin commercialization in Taiwan, leveraging established hospital partnerships, before expanding internationally.For Sahani and Prajapati, participating in the Best AI Award was less about winning and more about validation. They wanted to know whether their work mattered beyond the lab - whether people outside academia could see its value. Winning the Bronze Medal provided that answer.Prajapati admits they did not expect to win. The recognition, especially among international teams, gave them confidence that NeuroSpine AI is not only meaningful but scalable.Both researchers express strong interest in continuing their work in Taiwan, citing the strength of its AI, biomedical, and hospital ecosystems. For Sahani, the integration between technology and healthcare feels unusually seamless.Their roadmap is clear: expand anatomical coverage, refine clinical integration, and continue building AI systems with real-world medical impact.As Prajapati puts it, the most meaningful part of the journey has been having a platform to explain how AI can truly help surgeons - not in theory, but in practice. And in a field where millimeters matter, that distinction makes all the difference.NeuroSpine AI won the Bronze Award in the International Group AI Application Category at the 2025 Best AI Awards. If you have innovation would like to present, 2026 Best AI Awards with global tracks open for both AI Applications and IC Design, students and companies worldwide can compete for the grand prize of up to USD 30,000 (NTD 1,000,000). The deadline is March 16, 5:00pm (GMT+8). For more details, please follow official Linkedin for the lastest updates.
Friday 6 March 2026
Myogai Built with AI and designed for Humans
In today's wellness-driven world, yoga has evolved into more than just a physical practice - it has become a daily ritual for hundreds of millions of people worldwide. It now stands as one of the most widespread and diverse disciplines in the health and fitness industry.  Yet between "doing yoga" and "doing it right," a persistent instructional gap remains."In live classes," explained Myogai Project Manager Dale Neal, speaking from experience, "we often felt like we weren't getting the attention we needed. And instructors? They were struggling to manage larger groups while trying to keep instruction personalized." That experience became the spark behind Myogai.Led by Neal, Myogai is the product of a multidisciplinary team united at National Taiwan University of Science and Technology (NTUST). The group includes Nic, an expert in AI-based computer vision; Valer Vanco and Andrés Brítez, who bring strengths in business strategy and finance; Luis Manzanero, a full-stack developer; and Fatima, a seasoned yoga instructor with years of in-studio teaching experience.Together, they created a real-time yoga instruction platform that uses BlazePose pose tracking technology and intelligent AI feedback to assiststudents and instructors—enhancing accuracy, engagement, and safety.At the heart of Myogai is a multi-platform system that analyzes human posture in real time. Students perform asanas while the system tracks more than a dozen key skeletal points. Based on this data, the AI engine delivers instant alignment feedback, while instructors can access live dashboards that support more effective coaching - whether in physical studios or online sessions."It works both online and in-person," Neal said. "Our competitors often go to extremes - either they build AI tools that try to replace instructors completely, or they offer platforms too simplistic to be useful. We've built something in between. Myogai is a digital extension of yoga instruction - not a substitute."This hybrid approach is what makes Myogai a game-changer. By integrating real-time analytics into live or remote sessions, it allows instructors to teach more students effectively, deliver personalized feedback at scale, and even support AI-assisted certification programs.And yoga is just the beginning."Our roadmap includes everything from calisthenics to dance to physical therapy," Neal added. "Anywhere body movement matters, our system can make a difference."Myogai did not emerge from Silicon Valley, but from the startup ecosystem of Taiwan - a decision the team believes was pivotal."The startup environment here is incredible for AI," said Neal. "You get access to world-class engineers, rapid prototyping, and strong institutional support. NTUS's incubation program is what first connected us to this opportunity."Taiwan's technical talent, affordability, and proximity to leading hardware partners, such as sensor and device manufacturers - enabled Myogai to iterate rapidly and test early-stage concepts in real-world environments.Myogai's go-to-market strategy begins with independent yoga instructors - those running small studios or teaching virtually who need better tools to grow and retain their student base. With global organizations like Yoga Alliance listing over 60,000 instructors, the opportunity is substantial.The team also plans to expand into physical yoga studios, certification institutions, and fitness platforms. They are exploring potential partnerships with wearable technology companies to build richer, sensor-integrated experiences.Myogai recently won Silver at the Ministry of Economic Affairs'Best AI Awards - a major milestone that brought not only funding, but industry validation."This award shifted our mindset," Neal said. "We've always believed in our product, but now others in the AI community are recognizing its potential too. The visibility helped us land interviews, attract advisor support, and generate new business leads. It's already speeding up the rollout of our second version."That next-generation version, due to launch soon, will feature multilingual support, expanded device compatibility, and refined posture recognition models built on Mediapipe and enhanced with proprietary tuning.But for the team, this is just the beginning. The vision extends well beyond yoga - toward a future where AI doesn't replace human instruction, but elevates it across wellness, fitness, and rehabilitation."In a world rushing toward virtual everything, we're betting on something different," Neal said. "We're building AI that makes the physical world better—not obsolete."Myogai won the Silver Adward in the International Group AI Application Category at the 2025 Best AI Awards. If you have innovation would like to present, 2026 Best AI Awards with global tracks open for both AI Applications and IC Design, students and companies worldwide can compete for the grand prize of up to USD 30,000 (NTD 1,000,000). The deadline is March 16, 5:00pm (GMT+8). For more details, please follow official Linkedin for the lastest updates.
Tuesday 3 March 2026
ROHM Boosts GaN Supply with TSMC Technology
ROHM Co., Ltd. (hereinafter "ROHM") has decided to integrate its own development and manufacturing technologies for GaN power devices with the process technology of TSMC, with which ROHM has an ongoing partnership, to establish an end-to-end production system within the ROHM Group. By licensing TSMC GaN technology, ROHM will strengthen its supply capability to meet growing demand for GaN in applications such as AI servers and electric vehicles.GaN power devices offer excellent high-voltage and high-frequency performance, helping to improve efficiency and reduce size in a wide range of applications, and are already used in consumer products such as AC adapters. Adoption is also expanding in high-voltage applications such as power units for AI servers and on-board chargers for electric vehicles (EVs), and demand is expected to continue growing.ROHM began developing GaN power devices at an early stage and established a mass-production system for 150V GaN at ROHM Hamamatsu in March 2022. In the mid-power range, ROHM has built its supply structure while advancing external collaborations. One of the key partners in this effort has been TSMC: ROHM has adopted a 650V GaN process since 2023, and in December 2024, the two companies entered into a partnership related to automotive GaN, further deepening their collaboration.This latest integration represents an evolution of that partnership. Under a newly concluded license agreement, TSMC's process technology will be transferred to ROHM Hamamatsu. ROHM aims to establish the production system in 2027 to meet expanding demand in applications such as AI servers.Upon completion of the technology transfer, ROHM and TSMC will amicably conclude their automotive GaN partnership. At the same time, the two companies will continue to strengthen collaboration for higher efficiency and more compact power supply systems.