Imaging System Speeds Diagnosis with Real-Time Biopsy Analysis

Results of a pilot study conducted at the University of Rochester showed that a system based on two-photon fluorescence microscopy (TPFM) enabled rapid diagnosis of nonmelanoma skin cancer through real-time imaging of unprocessed, fresh tissue biopsies. TPFM imaging of nonmelanoma skin cancer was able to occur within minutes of obtaining biopsies, and the system provided histological features comparable to those of conventional histology.

The TPFM-based system could be used to diagnose skin cancer and other dermatologic conditions at the point of care. Patients could be treated immediately after diagnosis, rather than over a period of several weeks and multiple visits to the doctor.

TPFM uses near-infrared light, which penetrates deep through tissue, making it suitable for rapid imaging of fresh, irregularly shaped biopsies with minimal preparation. It can generate high-resolution, virtually H&E (hematoxylin and eosin)-stained images. TPFM with virtual staining can easily be performed at video rate, for multiple tissue specimens in parallel, and in real time, enabling rapid diagnosis.

In contrast, it takes several days from the time of a traditional skin biopsy to diagnosis. As a result, treatment is delayed, and additional clinic visits may be required for definitive care.

“This is how it has been done since the late 19th century,” professor Michael Giacomelli said. “Billions and billions of biopsies have been done this way. Everybody is on the same page; it works great. The problem is that it is very slow.”

In the clinical TPFM approach, researchers used a 1040-nm laser that scanned across the specimen and excited fluorescence at 16 f/ps. Emitted fluorescence from fluorophores was split and filtered into 518- to 558-nm and 620- to 680-nm bands. The two fluorescence channels were detected with two silicon photomultipliers that enabled high-speed, high signal-to-noise imaging, while being sensitive to signals from possible contaminants.

The microscope was self-contained in a light-tight enclosure, enabling imaging in a lighted room. The laser, microscope, controllers, and computer were integrated and mounted on a mobile cart.

To evaluate TPFM diagnostic performance compared to conventional histology, the researchers examined 29 freshly excised biopsies from confirmed nonmelanoma skin cancer lesions in patients presenting for treatment.

The team acquired digital TPFM images of the biopsies after excision and assessed them against their corresponding co-registered H&E slide images. Twelve of the co-registered image pairs were used as a training set. Fifteen were used in a masked evaluation by a board-certified dermatopathologist. Two pairs were excluded from the study.

When tested on 15 biopsies of known nonmelanoma skin cancer, the technology detected basal cell carcinoma with 100% sensitivity and specificity and squamous cell carcinoma with 89% sensitivity and 100% specificity.

The researchers further reported that TPFM could be used to histologically evaluate fresh tissue specimens faster —within 2 to 3 minutes — than frozen or paraffin sections, without the need for a histopathology laboratory or specialized personnel.

Though TPFM enables deeper imaging than other fluorescent imaging techniques, imaging is still restricted to approximately 100 μm into tissue.

However, as with conventional histology, specimens can be bisected or bread-loafed to expose internal tissue for imaging, eliminating the need for deeper imaging.

Tissue biopsied with a new imaging system based on two-photon fluorescence microscopy (TPFM). The system, which has shown promising results, was described in the journal JAMA Dermatology. It was developed by University of Rochester biomedical engineer Michael Giacomelli. Courtesy of the Giacomelli lab.

The prototype device is smaller than a cryotome, is portable, and requires substantially less operator training than standard tissue processing. This allows for real-time, point-of-care interpretation of skin biopsies, even in low-resource settings. TPFM imaging is nondestructive, and stains are removed by paraffinization; thus, it does not preclude subsequent histology or immunohistochemistry.

Giacomelli sees the potential for the system to quickly provide biopsy results for a range of diseases. For example, “in prostate cases, surgeons often have a very poor idea what they’re getting into based on just an x-ray or MRI image beforehand,” he said. “This technology offers a possibility to adjust the patient’s therapy on the fly.”

Study results suggest that TPFM has potential as a rapid, point-of-care diagnostic tool that requires no extensive sample preparation or retraining for image evaluation. Still, further validation of TPFM imaging in a larger cohort is necessary to fully evaluate its diagnostic accuracy. Giacomelli is working with professor Sherrif Ibrahim on a 200-patient follow-up study, using biopsy samples taken at random.

“The goal is to analyze how well the technique works in a real-world scenario, where you have a lot of people coming in and all kinds of things can show up,” Giacomelli said. “We really want to make sure that there isn’t some bizarre thing that normally has nothing to do with cancer that we somehow confuse with cancer — or something that does have something to do with cancer but we can’t image. You never know what you’re going to find once you start taking random biopsies of people.”

Giacomelli is also engaged in a parallel study to see if a new system he has developed incorporating TPFM and video can be used to guide surgeries. The system overlays TPFM images of the site where tissue is being removed on a webcam image, co-registering what it looks like to the eye with what it would look like processed on a slide.

“You can show what the excised tissue looks like, and how much of the tumor is located on the excision,” Giacomelli said. “They can then map that back into the wound that they are creating in order to decide which side to cut.”

The research was published in JAMA Dermatology (www.jamanetwork.com/journals/jamadermatology/fullarticle/2795725).