5-aminolevulinic acid is a natural precursor in haem synthesis. Given by mouth before surgery, it leads to accumulation of protoporphyrin IX in malignant glioma cells, which emit red light when illuminated in the violet-blue range (Stummer et al., 1998). After a prospective series showed that residual fluorescence predicted residual tumour (Stummer et al., 2000), a randomised multicentre trial found that fluorescence guided surgery produced more complete resections of contrast-enhancing tumour and better progression-free survival at six months than white light surgery (Stummer et al., 2006). The drug was subsequently licensed in Europe and later in the United States (Hadjipanayis and Stummer, 2019). Its positive predictive value for tumour is high, while weakly infiltrated tissue may not fluoresce visibly (Hadjipanayis et al., 2015).

Fluorescein was used to localise brain tumours as long ago as the 1940s (Moore et al., 1948). It is given intravenously, stays largely outside cells, and collects where the blood-brain barrier is defective, which in a high-grade glioma corresponds broadly to the region that enhances on imaging (Diaz et al., 2015). Shinoda et al. (2003) used a high dose that could be seen without any filter. The introduction of a microscope filter matched to the dye allowed much lower doses (Schebesch et al., 2013), and a multicentre phase II study reported a high rate of complete resection of enhancing tumour with this approach (Acerbi et al., 2018). No randomised comparison with white light surgery or with 5-aminolevulinic acid has established an effect on survival. The dye is cheap, widely available and has a long safety record in ophthalmology, though rare severe allergic reactions occur.

The barriers are practical. The licensed oral agent is costly per case and must be imported in many countries. Commercial fluorescence modules are tied to particular microscopes. Visual judgement of fluorescence is subjective and depends on ambient light and on the distance and angle of the microscope, as work on quantitative measurement has made clear (Valdés et al., 2011). Simple detection systems built from a light source and inexpensive filters have been described for fluorescein (Bongetta et al., 2016), which suggests that the optical requirement can be met cheaply. What such reports do not supply is the surrounding procedure that makes the technique safe and repeatable, and that is the gap the present workflow aims to fill.

Fluorescence guidance is most useful for intrinsic tumours that are expected to be high grade and for which a maximal safe resection is the aim. Where magnetic resonance imaging is not available, that judgement rests on contrast-enhanced computed tomography and the clinical picture, and the team should accept that some cases will prove to be other lesions. Fluorescein also accumulates in metastases, abscesses and other lesions with a disrupted barrier, so fluorescence does not confirm the diagnosis. Consent should cover the off-label status of the dye for this purpose, the possibility of allergic reaction, and the temporary yellow colouring of skin and urine. A history of reaction to fluorescein is a reason not to use it.

The dye, its expiry date and the resuscitation drugs needed to treat anaphylaxis are checked the day before. Dose and timing follow a single written local protocol based on a published regimen, agreed with the anaesthetist and the pharmacist, and are not varied from case to case. This article does not recommend a dose. The point of the stage is that timing determines contrast: fluorescein given too long before the margin is inspected spreads into oedematous brain around the tumour and loses its discrimination. The time of injection is therefore written on the theatre board. For 5-aminolevulinic acid the equivalent step is oral administration some hours before anaesthesia in line with the product licence, followed by protection of the patient from strong light for the period the licence specifies.

Before the patient is draped, the light source and filters are tested on a reference target, for example a card marked with a dilute drop of the dye beside an unmarked area. If the mark cannot be distinguished clearly at working distance, the case proceeds under white light alone. The theatre lights are arranged so that they can be dimmed or switched off during inspection, and windows are covered. The check takes a few minutes and can be added to the equipment section of the surgical safety checklist, a tool whose value in varied settings is well supported (Haynes et al., 2009).

The bulk of the tumour is removed under white light using ordinary microsurgical judgement. Filtered light is used at defined moments: on first exposure, to confirm that the lesion fluoresces; as the resection approaches its planned limit; and at the end, to inspect the cavity wall systematically. Fluorescence is treated as a prompt to look again and never as an instruction to resect. Anatomy, the proximity of eloquent cortex and tracts, and vascular territory set the limit, and fluorescent tissue in a functionally critical area is left. Blood and cottonoids are cleared before each inspection because both alter the signal.

For each case the team records the agent, the time of administration, the result of the optics check, whether residual fluorescence was seen at the end and why any was left, and any adverse event. A paper form is sufficient. Where the microscope has no camera, a photograph through the eyepiece is an acceptable record. Postoperative imaging, of whatever kind is available, is compared with the surgeon's impression. A monthly review of the forms shows whether the technique is being used consistently and whether the surgeon's reading of fluorescence agrees with imaging and histology.