Radiologist Detects 'Suspicious Asymmetry' in Crumb Matrix; Loaf Placed Under Follow-Up Scanning Protocol for 6 Weeks

At 11:43 pm on a Tuesday in October, Dr. Priya Nandakumar cross-sectioned Lot 14 and immediately flagged what she would later describe, in a calendar entry, as a "bimodal alveolar distribution requiring clinical correlation." The loaf was a 79% hydration country white, bulk-fermented for thirteen hours at 77.6°F with the 2:00 am stretch-and-fold omitted due to a trauma consult. Malpighi — her starter, maintained since 2021 and named for the Bolognese anatomist whose capillary microscopy work Dr. Nandakumar had cited in exactly zero radiology reports but thought about often — had passed all viability markers that morning: 2.9× rise at peak, pH 4.1 at the twelve-hour mark, strong bubble formation across the full surface of the jar. The crumb itself, measured against her seventeen-loaf baseline, was performing in the top quartile on every metric she tracked. The asymmetry she flagged — a relative alveolar density gradient running inferior-right to superior-left across what she had begun calling the coronal plane of the loaf — would not have been detectable without the Epson Perfection V600 flatbed scan protocol she had implemented in Q1. At 6400 dpi, the cross-section read like a chest film. Dr. Nandakumar, who had read somewhere between 40,000 and 60,000 chest films in her career at UCSF Medical Center, knew what she was looking at. "The inferior-right quadrant is hyperaerated relative to superior-left by approximately 15 to 18 percent," she told her husband Marcus, holding the laptop the way she holds a lightbox during case review. "That's not fermentation noise. That's a distribution." Marcus, a civil engineer who had been asleep, said he trusted her assessment. The follow-up scanning protocol she established before morning rounds was, in her own estimation, appropriately conservative for a first presentation. Lots 15 through 20 would each receive cross-sectional documentation at the 5mm, 10mm, and 15mm slice positions. She ordered a second Epson V600 — for comparison imaging, she explained to Marcus, in the same tone she used when explaining to residents why bilateral comparison views are non-negotiable — and downloaded the open-source ImageJ plugin she used for pulmonary density quantification, which she adapted over two evenings to accept crumb cross-sections as input. She named the adapted protocol BREAD-RADS. BREAD-RADS Category 3 — "Probably Benign; Short-Interval Follow-Up Recommended" — was where Lot 14 landed after a second read. This was, she noted in the Notion database she had built for crumb imaging, actually reassuring. Category 3 carried a malignancy rate below 2% in the clinical literature. She was not using the word malignancy here. She was using it structurally. "The architecture is sound," she told Malpighi on day two of the protocol, checking the starter's ambient temperature with the Inkbird IBS-TH2 Pro she kept clipped to the proofing shelf. "We just need serial imaging to establish a baseline." The six-week window was standard of care, she explained to Marcus that weekend, while laminating the BREAD-RADS decision tree she had color-coded by category. He was now responsible for the Saturday bulk fermentation log. She had sent him a shared Notion invite. He had accepted it, which she recorded as consent. Lot 20 came out of the oven on a Friday and showed the same gradient. Dr. Nandakumar documented this as "stable asymmetry, likely constitutional" and moved Malpighi to a whole-rye feeding substrate to assess whether flour composition was a confounding variable. She scheduled a follow-up read for six weeks. Marcus, who had begun photographing his own cross-sections for the shared database, submitted Lot 21 before she asked.