liveBook · Manning

Tumor

As explained in Applied Reinforcement Learning

A tumor is an abnormal mass of tissue formed by uncontrolled cell growth and targeted for destruction in radiotherapy.

A tumor is an abnormal mass of tissue formed by uncontrolled cell growth. In radiotherapy, the aim is to deliver enough radiation to control the tumor while limiting exposure to nearby healthy organs, known as organs at risk. section 8.4

Treatment is divided into fractions delivered over multiple sessions. Dose accumulates, and the tumor may shrink when treatment is effective or grow when the dose is insufficient. Beam direction and intensity must therefore be considered across the whole sequence, not just one session. section 8.4

Tumor in a treatment simulation

A treatment environment tracks cumulative tumor dose and tumor size alongside doses to organs at risk. Its simplified tumor model shrinks the tumor when the dose in a step meets a minimum threshold and otherwise lets it grow. The reward favors progress toward the target dose and penalizes tumor size and organ-limit violations. section 8.4

Listing 8.4 Updating tumor size based on delivered dose
def _update_tumor_dynamics(self, tumor_dose_this_step: float):
    if tumor_dose_this_step >= self.min_effective_daily_dose:  #A
        self.tumor_size *= (1.0 - self.tumor_shrink_rate)  #B
    else:
        self.tumor_size *= (1.0 + self.tumor_growth_rate)  #C
 
    self.tumor_size = np.clip(
        self.tumor_size, self.min_tumor_size, ➥self.max_tumor_size
    )  #D

The update illustrates the simulation’s threshold-based tumor growth and shrinkage rule. section 8.4

How Applied Reinforcement Learning covers it

How deep it goes
Explains radiation response and models tumor size with a simplified dose threshold rather than detailed tumor biology.
Examples
Models tumor dose and size in a fractionated-treatment simulation with organ-dose penalties and a target dose.
What you can do after
Evaluate/understand how tumor response and organ-dose constraints shape a simulated treatment-planning decision.
How it connects
Connects tumor response to radiotherapy, organs at risk, cumulative dose, and reinforcement learning.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Applied Reinforcement Learning. The section links point to the parts of the book it draws on.