Skip to content
OpenReadout

PerkinElmer Spotlight .fsm images

PerkinElmer Spotlight FT-IR imaging systems (with Spectrum IMAGE) write a map of spectra to one .fsm file. OpenReadout returns the spectra as one trace with a sweep per pixel, a table of pixel positions, and an image with one channel per spectral point. The instrument settings use the same blocks as .sp files.

Derived from specio’s FSM plugin (BSD-3-Clause, read as documentation and run as a reference reader), the .sp notes and hex dumps of two public files from two depositors (Spotlight with Spectrum One, and Spotlight 400 with Spectrum 3). See docs/provenance/perkinelmer-fsm.md.

Crate: openreadout-spectro, format id perkinelmer-fsm (FSM_FORMAT_ID), extension fsm, reader FsmReader; the dataset is the shared SpectroDataset.

Detection

PEPE, then a 40-byte description containing 4D (DataSet - 4DConst3DInterval) → definite. The .sp reader claims PEPE files whose description does not contain 4D.

Layout

After the 44 bytes, blocks: u16 id, i32 body size, body. Unknown blocks are skipped (listed in the vendor tree); a block past the end is a truncated finding.

  • 5100 — geometry: u16-length name (may be empty), ten f64 (x step, y step, z step, z first, z last, two 4-D limits, x origin, y origin, z origin), three i32 (width, height, points), four u16-length texts: the y and x stage-axis labels, the spectral unit (cm-1) and the value unit (%T, A, %R, KM).
  • 5104 — records: a block tree like an .sp data set: history records (121: user 35698, operation 35699, date 35700) and the instrument record (123: model 35837, serial 35838, firmware 35839, scans 35840, detector 35841, source 35842, beamsplitter 35843, resolution 35844, apodization 35845, spectrum type 35846, beam type 35847, phase correction 35849, accessory 35854; laser wavenumber 35882).
  • 5105 — one spectrum: points float32 values; one block per pixel, along x first.

Mapping

One trace (named by the image name, else <quantity> spectra) with one sweep per pixel, the x axis regular from z first to z last, the y quantity from the value unit as in .sp (transmittance %, absorbance AU, reflectance %, kubelka_munk); table positions (x_px, y_px, x_um, y_um: origin + index × step, µm); one image, width × height, one float channel per point, pixel size = the steps. Rows are in acquisition order (the stage y grows by one step per row); which way up Spectrum IMAGE draws them is not stated in the file.

Trace extra (our vocabulary): the .sp instrument keys (instrument, instrument_serial, instrument_firmware, detector, source, beamsplitter, apodization, spectrum_type, beam_type, phase_correction, accessory, scans, resolution_cm1, laser_wavenumber_cm1, acquired_at) and size_x, size_y, x_units_text, y_units_text, stage_axis_labels, origin_um, step_um. Image extra: row_order.

Experiment. instrument (vendor PerkinElmer, model, serial, firmware as software version), acquisition (started_at and operator from the oldest history record), sample.id (the image name), method.parameters: resolution, scans, laser_wavenumber, detector, source, beamsplitter, apodization, accessory, pixel_size.

check finding codes

truncated, block_size, spectrum_count (errors), x_interval (warning).

Validation

specio 0.1.0 on both corpus files: spectrum and point counts, six whole spectra bit for bit, the x axis (24 samples, 1e-9), the middle-band image bit for bit, model and serial.

Known gaps

Only the 4-D constant-interval layout has been seen. The value unit text can disagree with the spectrum type (%T on a Ratio (%R) image); the unit text is used and both are reported.

Vocabulary (public identifiers of the FSM module of openreadout-spectro)

identifier meaning
FsmReader, FSM_FORMAT_ID the reader and its format id

How this reader was derived, file by file: provenance log.