Biology in Action: Reducing the Human Cost with AI Image Analysis

Danielle Mandikian is excited by the impossible. As senior principal scientist at Genentech, she researches next generation therapeutics: the way a therapy is distributed in the body, and how it functions. But it’s not just the poetry of fluorescence microscopy that keeps her obsessively returning to her role. The human cost is always at the forefront of Mandikian’s mind. When working against the clock to find the right treatment for patients, Mandikian turns to microscopy and AI image analysis.

Danielle Mandikian never aspired to be a scientist. As a teen, she preferred welding and auto mechanics to school. Family disagreements over her future ambitions propelled her into community college where she discovered an abandoned microscope. It took her life down an unexpected road to California: to be closer to her family, and to begin her career as an academic. Her PhD marked a period of personal grief and professional breakthroughs. It also sparked an intense lifelong commitment to biotech research.

Towards the end of Mandikian’s PhD, the news of her dad’s terminal illness shattered her world. “I canceled everything: moved home”, she says. Confronted by astronomical medical bills for her father’s oncology treatments, she searched for a job that would support her family yet keep her moving towards her professional goals. Soon enough, Mandikian scored a job as a biotech researcher at Genentech, where today she is the Senior Principal Scientist. After standing in the patient’s shoes, she stepped into the role with a heightened level of empathy, excelling beyond basic and cell biology. In full circle, her work is inspired by the “major pivot” that changed her life.”

As a scientist, working against the clock to find the right treatment can create immense pressure and struggle. And for a functional medication to work, many stars must align. It’s not just the financial cost, but the human cost which is a priority for Mandikian. She’s aware that the funding which supports scientific research traces back to the taxpayer or “people who have lost someone they love.” Mandikian’s experience with her father solidified the notion that, in her words, “we're also spending the hope and desire from people.”

Mandikian’s daily work is now occupied with large molecule therapeutics: think, antibodies, peptides. She says her team is interested in how a therapy is distributed in the body “and how it functions when it gets there.” She describes her work as “multidisciplinary,” where her mission is to understand where therapeutics are distributed inside the body, and “to ask really complicated questions” at scale, and quickly. This translates into a faster turnaround time for therapeutics and helps her to predict the benefits to a patient.

The challenge with this approach, however, is that it’s impossible to know what is happening with the sub cells that are inside the tissue. Naturally, Mandikian’s group turned to microscopy to address the issue, where a whole new set of problems ensued. Bringing in new technologies to supplement their main system didn’t suffice, due to what she describes as the “hundreds of microscopy slides” they were faced with. The evolution of issues led Mandikian to invite an engineer from an entirely different scientific background to join her group.

Working on the scale of automation, the engineer sourced a microscope that could help them to process the hundreds of microscopy slides in bulk. “As soon as we accomplished that, we had way too many images to deal with” she says. It wasn’t just the sheer number of images that was the issue. The group’s questions were much more complex than what they could gather from just looking at an image. “That’s where AI really helps,” she says, “because it ensures that we can ask unbiased questions, in bulk.”

Mandikian used to find image analysis with automation intimidating – but she’s surprised at how accessible it’s become. Non-coders can find user-friendly software out there which breaks an image into pieces. She counts the many advantages of using AI: from its “sheer bulk force” to translating her questions from a biology perspective into an AI system.

AI now makes it possible to ask (and answer) more complicated biological questions than ever before. For Mandikian, that’s a huge deal, especially for health equity and the ability to distribute therapies that sustain temperature and other practicalities worldwide. But what does that mean for the future? Mandikian says she’s curious as to what will happen: “I think the real value is in accessibility for scientists.” Packaged software such as those offered by ZEISS, is bringing AI tools to a wider audience. Mandikian says that “The coolest thing about AI is finding different biological patterns that we would never have thought to ask for.” In that sense, consider it an application for understanding diseases.